Storytelling with Data & Data Visualization [2020] | Joshua Brindley | Skillshare

Playback Speed

  • 0.5x
  • 1x (Normal)
  • 1.25x
  • 1.5x
  • 2x

Storytelling with Data & Data Visualization [2020]

teacher avatar Joshua Brindley, Data Visualization Expert

Watch this class and thousands more

Get unlimited access to every class
Taught by industry leaders & working professionals
Topics include illustration, design, photography, and more

Watch this class and thousands more

Get unlimited access to every class
Taught by industry leaders & working professionals
Topics include illustration, design, photography, and more

Lessons in This Class

29 Lessons (5h 13m)
    • 1. Introduction

    • 2. Speaking the Language of Data

    • 3. Effective Data Communications

    • 4. Project Outline

    • 5. What Makes and Effective Data Communication

    • 6. Effective Data Communication Examples

    • 7. Visual Perception - Order

    • 8. Visual Perception - Hierarchy

    • 9. Visual Perception - Clarity

    • 10. Visual Perception - Relationships

    • 11. Visual Perception - Convention

    • 12. Visual Design and the Application to Data Vizualisation

    • 13. Components of a Data Vizualisation

    • 14. Different Types of Graph

    • 15. Deadly Sins of Graph Design

    • 16. How to Avoid being Mislead with Data

    • 17. Remove Distracting Elements

    • 18. Bringing out the Story with Colour

    • 19. Analytics Value Chain

    • 20. Understanding Context

    • 21. Tales from Work: The Importance of Context

    • 22. Data Narratives with TEMPLATES

    • 23. Turning a Graph into a Story

    • 24. 6 Steps to Telling a Data Story

    • 25. Thank You

    • 26. CASE STUDY 1 - Overwhelming Monthly Report into a Succinct Data Story

    • 27. CASE STUDY 2 - Conflicting Data into Compelling Direct Narrative

    • 28. Project Submission

    • 29. Project Answer

  • --
  • Beginner level
  • Intermediate level
  • Advanced level
  • All levels
  • Beg/Int level
  • Int/Adv level

Community Generated

The level is determined by a majority opinion of students who have reviewed this class. The teacher's recommendation is shown until at least 5 student responses are collected.





About This Class


These days, the world runs on data. Over 2.6 quadrillion bytes of data are produced every single day and data visualization has emerged as the common language.

People in all levels of business are expected to tell stories with data if they are to participate in a world driven by data.

This course provides you will all the skills necessary to tell insightful data stories that drive action and resonate with an audience.

We will learn how communicating with data is important in today's world, the role context plays in telling data stories, how visual design and visual perception support data visualization, and how to turn your default graphs into impactful data visualization stories.

Meet Your Teacher

Teacher Profile Image

Joshua Brindley

Data Visualization Expert


Hello, I'm Joshua. 

I',m passionate about data, and even more passionate about sharing.


I love telling great stories with data, and driving results with visualisation that resonate with an audience.


I've helped plenty of people learn that art of communicating in the language of data, and I'd love to help you too!


Visual Design, Storytelling Skills, Understanding the Context of Analysis, Formatting Graphs and Data Narrative templates is what i teach in a fun, engaging way to teach you how to tell impactful stories with your data.

See full profile

Class Ratings

Expectations Met?
  • Exceeded!
  • Yes
  • Somewhat
  • Not really
Reviews Archive

In October 2018, we updated our review system to improve the way we collect feedback. Below are the reviews written before that update.

Why Join Skillshare?

Take award-winning Skillshare Original Classes

Each class has short lessons, hands-on projects

Your membership supports Skillshare teachers

Learn From Anywhere

Take classes on the go with the Skillshare app. Stream or download to watch on the plane, the subway, or wherever you learn best.


1. Introduction: Hi there and thank you for your interest in this cost. My name is Joshua bread Lee and I'm a data analyst and an analytics instructive with a particular passion, an area of focus on data visualization and visual communication with data. Discovering and sharing insights from data is what I do. And sharing is where I'm at my best. In this course, storytelling with data visualization, you will learn how to present data insights to a captivated audience to drive actionable change. You will learn how to create beautiful and stunning data visualizations that will inspire an audience into action by clearly presenting an inflammation of effective charts and graphs. In this course, we're going to explore the exciting world of visual communication and visual perception theory to understand the fundamentals of what it means to communicate visually and how intrinsically as humans, we interpret meaning from visuals. Then we will borrow lessons, visual design as in marketing, movie on the odds industry to learn how the experts communicate and inspire with visuals and graphics and how all of these lessons can be applied to the weld of data visualization in order to tell your data-driven stories will also explore the fascinating field of ethical data visualization and discover how accurate and truthful data and information can be presented in a way that it's misleading at dishonest or as I like to say, line with the truth through data visualization and rule and practical ways to avoid these designed features. Your data visualizations are accurate and represent data truthfully. So if this sounds interesting, then you'll love progressing through this costs, which contains over five hours of high-quality video, unfold cola Cost companion guide, job aids that you can download and can't forever. And once you've reached the end of this course and progress through the quizzes and knowledge checks. Then you'll have case studies and user submitted assignment that I will provide 101 feet high quality. So what are you waiting for? Enroll now and lend to oxidize the odds of data visualization. I look forward to seeing you in the course. 2. Speaking the Language of Data: hi there. These days the world runs on data over three quadrillion bytes of data is produced every single day on that pace in which we create data is accelerating rapidly as more people in things generate data. With the increase of data, generation has emerged the language of data to help businesses communicate with each of it in this new medium. So what is the language of data? How do we communicate with it and what is it used for? This data isn't just collected and starred so that it can be forgotten. It is used by businesses to make decisions and inform their thinking. Every single decision made, no matter how big or how small, requires data to inform the decision. Take a Monday, in example, such as choosing a hotel. The data you collect to inform that decision would be to look at the location of the hotel , the cost, the facilities, the amenities. And then you compare that to the budget that you have, but the amenities you want, what facilities you want in the hotel. You might also consider how long you need to be at the hotel based on your travel itinerary . This is old data that you collected to inform a data driven decision on what hotel you want to select in the world of business, this kind of decision making is done every single day at all. Levels of business data is the new lifeblood of business, and visualization has emerged as the common language. Effective communication with data is now a must have skill for all employees at all levels . Every single person in the organization is expected to speak this new language if they are to participate in a world driven by data. Think about how often you encounter a data communication in your life. Perhaps you have a fitness tracker app. That visualize is this step you've taken calories consumed or weight lost news sites that you visit use visualization all the time to communicate the news. Your social media also uses visualization to show your engagement history of the app or friends in the workplace. You've probably encountered many graphs, such as project Gant, charts or status or parts. All of these examples are communicating in the language of data. In order to be a part of the workplace, you must be able to speak this language companies are so relying on data today that being able to confidently work your spreadsheet and click on the graph creator bottom and leave it on the basic default graph is no longer going to cut it. The employees now needs to be able to create impactful and meaningful charts that resonate with an audience. The data and information could both be correct between two employees but the employees that can create a clear, accessible message. We've proper formatted data visualization will be ahead of the employees that cons. So speaking this language isn't about learning numbers, math or statistics. It's more about visual thinking and storytelling skills applied to graphs and charts. So I'll see you in the next video, where we reviewed the role that proper data communication, please on what qualifies as an effective data story. 3. Effective Data Communications: Hi that speaking the language of data, it's the intersection between data visualization. On good practice were visual and design thinking. A graph can be analysed, provide insight to the business and drive decisions with data. But a standard Barta doesn't really say a whole lot to an audience, at least one that isn't particularly analytical. So how can we enhance that bar chart in order for it to resonate with your audience? A data visualisation needs to be enhanced if it's gonna resonate with an audience. So how do we create impactful data visualizations? To do so, we must first take a step back and ask What is a data? Communication data Visualization is an encompassing term that refers to the techniques used to communicate data and information to an audience by use of visual objects such as the points lines bars on a graph, it's simply a visual display of data. Any instance where you take data on display visually, such as a graph, is a data visualisation. Visualizations can vary in complexity from this deep, complex graph to something more like this graph, or sometimes simply just a number. A graph is simply display of numbers. It is only when those numbers are looked up and interpreted with the proper business context, with the purpose off in forming a decision. Does that graph tell a story? So if a graph is just a data visualisation, a data communication is when a single or multiple visualizations air used along with visual communication elements to convey a message. Data communications can take many forms, such as reports, dashboards, infographics or presentations on the goal of a data. Visualisation is to clearly and effectively communicate information to an audience so they can take a decision. Vitaly Friedman, the founder of Smashing magazine and Design lover, put it best in 2000 when he said the main goal of data visualisation Issa, communicate information clearly and effectively through graphical means. It doesn't mean that data visualization needs to look borrowing to be functional or extremely sophisticated toe look beautiful. To convey ideas effectively. Both aesthetic farm on functionality need to go hand in hand, providing insights into Arriva sparse and complex data set by communicating its key suspects in a more intuitive way. Yet designers often fail to achieve a balance between form and function, creating gorgeous data visualizations, which failed to serve that main purpose. To communicate information, Friedman touched on something important here to achieve a balance between form and function . Graphs can be accurate and contain correct data, but these are just numbers that don't say all too much to an audience. So because they are designed well, they failed to achieve their purpose to communicate information. Similarly, you can have a really stunning and beautiful graph that doesn't really say all too much to the viewer about that business or really lead to an action or decision being made. So striking. The right balance is key. An effective data communication is one that is clear, easy to understand. Andi has an impact on the viewer. It's one that tells a story. In his 1983 book, The Visual Display of Quantitative Information, Edward Tufte defines graphical displays and principles for effective graphical display in the following passage. Excellence in statistical graphics consists of complex ideas communicated to clarity, precision and efficiency. Graphical displays should show the data introduced the viewer to think about the substance rather than about methodology, graphic design or the technology of graphic production or something else. Avoid the starting what the data has to say Present many numbers in a small space. Make large data sets coherent. Encourage the I to compare different pieces of data. Revealed the data. Several levels of detail, from a broad overview to the fine structure. Serve a reasonably clear purpose description, exploration, tabulation or decoration. Be closely integrated with the statistical and verbal descriptions of a data set. So without the proper consideration and application of these principles, your visualizations will fail at their purpose. It's incredibly easy to create an ineffective graph. Tools such as Excel Google sheets make it so easy to create graphs. With just one click, you can create any graph e one, but their basic, the default graph they create doesn't follow any other principles laid out by tuft and so that ineffective at communicating data to an audience. So join me now in the next section of the costs, where we will review both good unde bad examples of graphs, demonstrate these principles in action and understand what makes some graft effective on why offers failed to achieve their purpose. I'll see that 4. Project Outline: Hi there. So let me introduce you to the project. The project should be done twice in a sort off before and after fashion. The project provides you with a scenario and some data and asks you to find the answer on Put together a data driven communication for the management that tasked you with looking at the data. I encourage you to do this project now as a sort of first draft. Then make a copy of it and I should go through the cost, turned back to your copy of the project and make any updates as you learn them and start learning more about storytelling with data. Then at the very end of this cost, you can submit both your before on after to see how far you've truly come in communicating impact Feli with data. The very last video in this course is a review of the project that I did so that you can see an example and compare it to your own. However, I'm actively reviewing projects submitted and will happily provide you with feedback directly on your project. Also follow me on skill share where I share great examples of data storytelling. Highlight projects that I think are amazing as well as sharing data stories that I tell in the real world with step by step guides that will give you some inspiration to tell your data stories. So now let me share View the project brief for your first attempt. So now I want you to take on this activity. This is your opportunity to take everything you've learned and put together a compelling, rich data narrative. The brief is pretty simple. Lightspeed Cinemas operates three cinemas in one particular region, and they've touched you with looking at the data for the last three months on putting together a story that answers the question. Which of the three cinemas have the most successful concession stand? So you've been provided with a date, except that shows you the products each of these cinemas cell at the concession stand, the number of sales they generated, the revenue, the cost, each of the cinemas charges for these three products as well as the number of tickets sold . A little hint to you is to take a look carefully through the context so that you can understand what comprises a successful concession. Stunned in the activity you'll find the data set so you can use whatever tool you're comfortable. Way of toe work on this, and you have to put together a couple different graphs, perhaps on a presentation slide. A word document. Pdf. It's up to you entirely, but my tip to you remember, look through the context and really identify what it is that makes a successful concession stand. Good luck. 5. What Makes and Effective Data Communication: hi there. As companies become more relying on data, the number of people speaking the language in data has increased on with all these participants in this language, they begin creating a lot more graphs on. Unfortunately, not all of these graphs are effective. There are plenty of ineffective grafts out there, especially in the workplace. In this section of the costs, I'm going to share some knowledge of you that is both a blessing on somewhat of a curse. In order to create effective, impactful data communications, you're gonna have to learn how to spot when they are ineffective. So in this section, I'm going to show you how to identify ineffective graphs so that you can identify the key components that make them full shot of effective. But once you've learned this information, you're going to be able to spot any ineffective graph on. Unfortunately, you're going to start seeing them everywhere. Take a look at this example graph and ask yourself, Is it effective? Does it follow the principles laid out earlier by tuft? As you look at this graph, ask yourself what is the message or story here? Does it come across clearly, or do you have to interpret the graph and in for the intent of the offer about what the data is telling you. So is it effective? Well, no, it's cluttered, It's messy. It's very busy. There's a lot going on in this graph, and it distracts from the office intended message, making the communication unclear and ineffective. Consider what the intended message of this visualization is. There isn't really such an obvious one. By looking at the visualization, it could be saying a lot of different things, such as Q four was a relatively steady quarter of sales. Or perhaps the author intended to communicate. That Q one was a very varying sales period. Well, these are just guesses. The intent is not clear. So as an audience member, what we're doing is analyzing the visualization and trying to infer a message which isn't what you want to happen. Consider what Friedman said was the intent of a visualization to communicate a message. The data and the figures are accurate and true, but because of the presentation of the information, the graph fails to convey a clear message, meaning it won't have an impact on the audience. Now, take a look at this graph. This graph represents the exact same data as the one before, but this time we've removed all of the graph elements that distract the viewer and stripped it down to just the components that convey information. So now that the clutter has been removed, I ask again, Is this an effective graph? The graph is smart, easy to understand and coherent. So does that mean this graph is effective once again? Does this graph tell a compelling and clear story? Can you tell the author's intent? The answer is no. We cleaned up the graph, removed the distracting elements, but it still doesn't tell a story, which highlights the fact that data communications require more than a well presented graph . With the increase in the number of people who are fluent in the language of data, more and more graphs of being created in the workplace there are far more graphs that are impactful, persuasive and poignant. So graph like this simply fall short of effective. To be able to converse in the language of data is more than just a tidy graph. It's about communicating the story with impact. Think back to what Friedman said. Designers often failed to achieve a balance between form and function, creating gorgeous data visualizations, which failed to serve their main purpose to communicate information without the proper context. This graph alone doesn't fulfill that purpose. It presents data well, but it doesn't communicate a story. So this isn't an effective graph. So what I would like you to do now is a short exercise. Pause the video and jot down what it would take to understand this graph. Put yourself in the shoes of the sales manager having a monthly meeting on sales. And this was the graph there was presented the present assays. Here are the monthly sales and shows you this graph. Pause the video and think about what additional questions you would want to ask of the author toe. Understand what the point of this graph is? Okay, welcome back. You probably had a lot of questions you could have asked, such as other sales. Good. Why was the previous month so low? All sales continuing to rise. What is a seasonal sales period? Ah, well presented graph shouldn't leave the audience with so many questions about what it is that the author is trying to say. So Now take a look at this good example. Now look at this visualization. This data is the same as the previous graphs, but we've introduced some visual design elements that reinforces story of the visualization , coloring, spacing, size and text has been used to craft a narrative. In this example. The graph and the numbers don't stunned alone. The numbers now actors reinforcement to the narrative that is being presented to the viewer without being a part of this organization. You can instantly understand this story that's going on here. The organization had a period of continuous growth then, dudes economic factors. They had a dip of 589 or 5% in sales, which by March had recovered. This is an example of effective communication with data. This highlights that two equally important aspect of an impactful data communication. The only do you need a graph or grabs that are comprehensive and clear. You also need elements of visual design to create an effective narrative. So join me now in the next section where we will take a look at several good and bad examples so you can spot ineffective data communications in your workplace. 6. Effective Data Communication Examples: Hi there in this section of the cars were going to review some graph examples on identify the components that make them effective or ineffective at communicating with data. Well, look how we can create impactful data communications in the preceding sections of the costs . But right now, we're just gonna take a look at a few examples, so you can begin to identify the ingredients that make an impactful communication or what fell short in an ineffective one. So let's take a look at some examples and understand whether or not they are effective on what needs to be done to make them an impactful data communication. As we look through each example, ask yourself if it strikes the right balance between a clear, understandable graph on one that communicates a story. In this example here, the manager of a coal center has approached Hate Jar to make the case to hire additional staff. Take a moment to review this state of communication. Does the author communicate effectively the reasons why the manager wants to hire additional stuff? Let's review this data communication. The author starts of a bold title that explains what these numbers are telling us. Need additional staff. Then they provided some comments that read. As the number of calls increase customer feedback decreases had an increasing calls from September 2019. Need additional staff to maintain high customer feedback? Andi. They provided a table that shows the mumps number of calls and average feedback rating, which we can see from September. The number of calls increases, and so the average feedback rating decreases. This communication clearly lays out the reasons why more stuff should be hired. It's evident that as the number of calls the call center handles increases, the quality of the calls decreases. This communication also shows the impact of the business if they don't hire more stuff in terms of communicating information. This data visualization definitely takes the box. However, it doesn't do so particularly effectively. So let's take a look at the ways we can review this graph. With this particular communication, there is a narrative being communicated to the audience and the numbers A used to reinforce that message. However, they're just that numbers. The graph itself was not an effective choice of visualization. It contains a lot of irrelevant data, and the design could be improved from some of the visual design principles. We will go into far more detail about creating effective data communications and impactful graphs in the preceding chapters of this cost. But for now, I'm just gonna highlight good and bad aspects of graphs so that you can begin identifying them in your workplace or with the graphs that you create. With that in mind, let's take a brief look how we can improve their states communication. Firstly, let's remove the irrelevant data points. The audience doesn't need this many months of data to understand the key message. We only need enough data points to highlight. The average rating was high in the previous months before the Cole Center experienced its increasing calls, which lowered the average feedback rating. Next, instead of a table, we're going to replace it with a combo char that contains to Siri's. This is a far better choice of visualization as it communicates the message far more quickly. The visual nature of this graph encourages the I to compare. Plus visually like this, it's easier to pick out the trends. Finally, we're going to apply some design thinking principles to this visualization, which again we'll cover in a lot more detail in the preceding chapters. So take a moment to review this graph and sort of see the changes and how they've taken form. Now we've created an effective communication. It both communicates a clear story until so effectively, every element of the visualization, from the color to the labels on the written words add to the same story. So let's take a look at a second example. In this example, a 19 manager works for a large organization that mainly works on internal projects. The projects the I T team works on recharges. The resource is back to the organization. The past couple of months, his team have worked on four projects, and the Finance Department are confused as to why Projects four on one, despite having virtually the same amount of chargeable hours, are being charged at such different amounts. The I T manager uses this communication to explain why you compose the video. If you need a moment to sort of review what's in this communication, So is it clear and understandable? Does the visualization contain too many distracting elements? Can you point out what the reason Project one on four costs such different amounts? Well, The reason is actually because project fall was more complex and required different resources that charge higher rates and those on Project One now this example of data communication contain many of the right elements. It had the clear, readable graph that was easy to understand, but only one she knew what it was saying. The communication also contained text and over elements, but they all weren't contributing to the story. A good and effective data communication that will resonate with an audience contains all of these elements, such as a well presented data visualization supported by some graphical and text elements. But all these components must be telling the same story over wise. They begin distracting from it if they on concisely all singing the same tune than you just end up with a musical mess. So how can we change this communication to be more effective? Well, firstly, the title, the reason this communication exists on what it's there to communicate is that project for had more expensive resource is than Project One. So let's use that as the title. Now we're going to remove the irrelevant data points. We don't need projects two and three, so I'm gonna cut them out. Finally, we're gonna use over visual elements like the text and graphics to add to the story. So this is the revised version straightaway. Finance Department is being told the reason why, and everything else supports that. We can see the hours of each of the resource is that they put into the project. And here the cost of the resource is between both projects. So it's clear as to the reason why each element is being used to reinforce the narrative. And that narrative was spelled out clearly in this section of the costs. We took a look at a couple examples off data communications that are typical to a workplace . We reviewed each of them and highlighted why they weren't effective and demonstrated how you could make them more effective. So join me now in the next section, where we will take a look visual communication on the tools of visual designers used to tell compelling narratives with graphics and how we can use those tools when creating data communications 7. Visual Perception - Order: hi there. So so far we've looked at why speaking the language of data is an important skill in today's world. We learned what an effective data communication is on how to identify the components, often effective communication. We also know what makes an ineffective communication. So now, in this part of the costs, let's discuss visual communication to make impactful data communications. We've got to borrow some tools from visual design and visual perception practices. So what are these tools and how do they apply to data visualization? Communicating visually in the language of data is a lot like cooking. Okay, right. Hey me out on this one. Anyone can eat a meal and farm an opinion about it without having to take a course in food science, we can eat a meal and know whether or not we liked it. We may even be able to describe some aspects of the meal that we enjoyed, such as that it had a richness or a texture that we liked without knowing that what you're actually doing is describing the male yard reaction of the amino acids. Similarly, with data visualization, you can see a graph or data communication and just sense if it was effective or if it resonated with you, you could possibly even describe components that you liked, such as it was insightful or informative. And you can do this without having a degree in visual perception. If you wanted to pursue a career it a chef, you'd have to have at least a functional knowledge of food science equally. If you want to create data communications that drive decisions an impressive audience, you should probably know a little something about visual perception. So let's take a look at the broad concepts and learn how they can apply to creating impactful graphs. Firstly, let's talk about visual perception. Visual perception is the ability to interpret the surrounding environment using light in the visible spectrum reflected by objects in the environment. Okay, a bit nerdy that one. So simply put, visual perception is how we interpret visual images. It's about how we, as humans on our brain, processes images and turns colored shapes into understanding emotion and ideas. There's entire feels of research on this topic from multiple disciplines such as medicine, psychology, child development and graphical design. But when it comes to the application to data visualization There are five main concepts that you need to know order, hierarchy, clarity, relationships and convention. So now let's explore what each of these mean and how you can apply them to your data. Visualize Asians order. Visual perception is unlike text in text, there's an agreed upon standard in the order in which we read. The standard may differ between cultures, but in Western culture, we read from left to right top to bottom, just like in this image. If I flipped the words around so that they're now read from top to bottom, left to right, it's completely unreadable. It's basically a complete mess. That's because we expect the letters and the words to be in a certain order on when they aren't in the order. It's far more difficult to read, even though the letters are the same, and they're put together in the same way to create words. This passage is really challenging to read, regardless of culture, the written word conforms to standards. If you bought the latest Stephen King novel from a bookstore in America, you'd be surprised if the words are printed from bottom to top. But when it comes to visual perception, there is no agreed standard by which we follow, even in data visualization. So I have a little exercise for you. Look at this. Typical data visualization. There's some space in your workbook underneath this graph. I want you to write down the objects of the visualization in the order that you look at them. For example, I look to the legend first and then the access post a video now and complete this small exercise. Okay, so we're back. You have no way to confirm this, but just trust me on this one. I've done this activity many times, and every single one of you would have put down a different order in which you viewed the elements. This is my order. Yours probably looks different. It's good to be aware of the order in which viewers interpret your graphs because you need to help them interpret graphs in an efficient way by making correct design choices. There's no right order to view this in. Each viewer will read and understand this visualization in a different order. In fact, most people wouldn't have even read the title first. The way data visualizations, air read and understood is sporadic. The eye jumps around the chart, darting between different components as you did the activity, you probably noticed that the Saudi red them in isn't particularly structured. It doesn't go from top to bottom left to right like it does in writing. Each person will read in a different order. Andi, at a different pace, is their eyes scan across the visualization and come to an understanding about what it means. So how does this relate to data visualization? Well, the composition of the chart and even the charts election can impact the way your audience views and understands the message. What you need to know is that you can design your visualizations toe, lead the audience around it and help guide them to an understanding of the message which leads us to our second concept. 8. Visual Perception - Hierarchy: high rocky. So we know that when it comes to visual perception, the audience scans across an image in no particular order. Well, actually, the order isn't entirely random. Our eyes are drawn toward ever stands out. First, we tend to look directly. Whatever grabs our attention when it comes to data visualization, that is usually the bright colors, outliers, largest difference or most dense cluster. Review this visualization again. Whilst the order in which most people interpret this visualization would be different, most people would begin the lines themselves, and then they'd noticed the sharp increase for nonfiction in March and then move on to the legend. Perhaps your exercise looks something like that. This was a very intentional design choice. I selected dull, muted graze for most of the graph elements and chose bright colors that stand out for the data points. I was manipulating your perception with ease design choices to guide your understanding of the chart without you being consciously aware of what I was doing. All well designed data visualizations make conscious and deliberate color choices to draw the audience attention to what the author intends. But most visualization tools do not understand context and they don't understand what is actually being communicated so often. When you use a basic default chart and a tool such as Excel, the attention has been applied in the wrong places. For example, look at this graph. This visualization draws attention to the access and title, pushing the actual visualisation and data to the background. It stays the same messages, the previous graph, but this time it's a lot harder to interpret. Another way of poor use of this concept is to compete for the attention of the audience. Everything in this visualization wants your attention. So competes for your understanding. This confuses the message, and it makes it hard to understand the right way to use this concept is to draw the audience's attention to the focus or story of the visualization, to guide their understanding, not distract from it. Take a look at this visualization. For example, it shows the revenue generated for a cafe post a video if you need to look over it quickly in your workbook, you probably looked at the elements in this order, the line, the sharp drop, and then the competitors line. Then, in any particular order, you scanned over the rest of the visualization. Consider how the design choices have told you a story. You probably understood that this cafe was generating around 17,000 in revenue, then a competitive opened and sales talk a large dip. This is good use of focusing attention to an audience so that they can understand a story. So using bright colors, a muted graze to highlight attention is a really powerful way to focus your audience in on this story. So powerful, in fact, that it can actually hide the rial story that's within the data. You have to ensure you're focusing on the right message. Your design choices should be deliberate, not just generically using bright colors on the data elements and muted grace for the background. Because if you do that, you risk hiding the true story. I bet you miss the real story in this data because of the design choices. Consider this same graph. But the story hey, wasn't that the competitors opened and then sales dipped wife. The focus was on the overall continued growth this cafe has experienced over the cost of the last 11 months. You probably miss that because the focus was applied to the fact, a competitive opened. If you really look, you can see the growth story in there but isn't what you were drawn to. Understand. These design choices focus on a story but not the right one. The design choices must all contribute and support the same story. Otherwise, your message conflicts of your design, modeling the story and confusing your audience. So how can this peasant change the visualization To tell the right story? Consider this revised graph. The design choices now complement the story Riva than distract from it the title. Even though it is in the first place, people look as a clue to the understanding of the data visualization. The use of cola has drawn the attention to the right places. The part of the line that grabbed the attention, and now the upward trend and the parts that don't support this have been muted. Additionally, important context has been added, but the addition of the temporary slump and growth continues lines. This graph, with a few simple design Joyce is now tells a very different story, despite the data being exactly the same. So it's important that when you create your data visualizations, you aren't just using a generic formula to highlight attention, such as focusing the data elements and meeting the graph elements because you may end up inadvertently telling the wrong story. 9. Visual Perception - Clarity: clarity our brains of the apex of complexity. Trillions of neurons work together to make sense of the world phone consciousness. Andi, as a species rise to the top of the ladder. However, when it comes to visual understanding, our brains can only handle a surprisingly low capacity. You should know that your audience is only able to hold on, understand a few pieces of information at once. So when it comes to data visualization, simpler truly is better. So for an audience member viewing this graph despite there being about 20 different lines and four different categories meaning 80 different individual pieces of information, our brains are only going to understand two of them. Even if we wanted to understand more, we couldn't possibly. We're only gonna understand a general downward trend on one outlier book that's doing a bit better than the rest. This is because our brain seek clarity. Even if the audience member tried to hold on retain all this information, they're not gonna be able to know and understand 80 different data points. This idea of being able to understand the trend from many different data points could be expanded to include millions of data points taken extreme example like this one. This graph plots over 15,000 data points. Our minds don't even attempt understand any individual data point. They're drawn to a trend. What we can quite easily see from this graph is as the number of entities analyst ical increases, so did alikes, a mostly Stickles contained between 10 and 20 entities. So you can see how our brains are actually quite good at finding a trend in a data set. We don't view this graph as 15,000 individual data points. We actually view it as one or two. However, that isn't to say that you can just throw a lot of data points into a visualization on our minds will just somehow make sense of it. You must still consider proper design to demonstrate. Here we have the same data from the bookstore, but now you can't make heads nor tails of it. There's over 19 categories per bar. All colors compete for attention, so non stunned out. Some colors are so similar that you can't tell them apart and some are so small you can barely notice them in one of the bars. Poor design doesn't succinctly aggregate large data sets into an understandable trend. Nor does it allow for any individual data points to be understood in context. It confuses models the story and doesn't communicate effectively. So when it comes to the visual perception concept of clarity, it shows itself in two ways. In data visualization. So we just covered the 1st 1 That is, when we're presented with a lot of data. We don't pick out any individual data points. We see it as a whole. As a trend, we aggregate all of that data together, and then we actively pick out two or three features of that trend, such as it's a steep incline or a sharp drop. We don't pick out all of the information, but the second and possibly most important way shows itself in data. Visualization is how your audience actively participates in viewing your graph. Whenever I present a graph such as this plane standard graph, the audience is actually looking at it, but they're not taking everything in individually and then picking out on actively thinking and consciously being aware of all the data on it. What they're actually doing is picking out two or three bits of information and then everything else gets put into the shot term inactive memory, which is important for two reasons. One, when they leave the presentation or they leave the meeting or they read over the report that I've sent out and then they think back to it at a later time. They're only gonna remember a couple pieces of information. They won't remember all the individual bars, all the individual labels, all the different values. They'll just remember the key theme of it, which is important so that you present your graphs with that key theme in the forefront of their mind whenever you share them a graph. Remember the perception off order? They're actually not taking it all in there looking individual bars, individual elements of the graph one by one in a random order, usually on. Then they're only gonna remember the 1st 2 or three things they see. So you need to make conscious design and formatting choices so that the two or three first things they notice are the key themes or story off this graph. So take a look at this revised version. Now in the forefront is the story. I've brought it out using highlighting and focusing, and I've even added data elements and tax directly onto the graph to really bring the story out. So now when they view this graph, all the audience is actively participating and remembering the story and then at a later date when they will call it. They were calling the story and not just a random piece of the graph that they happened to look at first. So don't worry. We're going to go into far more detail about formatting and visualization design to bring out the story. For now, I just want to highlight the fact that you need to put this on your graph, Otherwise it's not going to resonate with the audience. 10. Visual Perception - Relationships: relationships. Our brains are wired to make sense and interpret meaning. Whenever you see something your hard coded to understand what you're looking at in the context of its environment and try and fit it into some sort of narrative, consider once more the buzzfeed list ical visualization. You looked at this, you understood the data, and you took away that most lis tickles contained between 10 and 20 items. Then, as you understood this, you start forming a story or narrative out of it. You start looking for the why in the data, it's instinct to try and put two and two together. Most people of you, this graph and then is thinks something along the lines off. Well, people probably just don't read articles that a to shop. They don't see them as having enough substance. And people don't finish articles that are too long. So the sweet spot is probably between 10 and 20. The data, it doesn't actually say anything like this, but people will form that story instinctively, probably unaware they've even done so. This is important when it comes to data visualization, any audience viewing the chart will create a narrative. They begin to understand the data. So the order in which you show the data is crucial on this story in which you tell should be directive. So now I want you to think back to the cafe. As you looked over this graph, the first thing you notice is the sales, then the depth than the competitive line, and you start putting two and two together and instinctively create a story out of that. You probably thought something like a competing cafe opened. Customers began going to the new cafe. They mustn't have liked it. And then they came back to this cafe. It's instinctual to seek a narrative out of the understanding the visualization like that. Now suppose that in the example of the cafe, the reason for the sales dip was actually because they were refurbishing part of the cafe, so they had less available seating and could therefore do less sales. While it is true that a competitors opened at the same time, it didn't actually have any impact on the cafes revenue. But because it was presented you as an audience member, created a narrative out of it, thinking that the competitors had nothing to do with the sales dip didn't even enter into your thought process because, well, why would it? It was presented. Therefore, it must be part of the narrative, right? You might be familiar with the phrase Correlation is not causation. This phrase exist to remind ourselves that just because something is presented together doesn't actually mean they're related. Well, regardless of this phrases existence, your audience will instinctively create a narrative out of whatever is presented. Okay, so sticking a label on a graph that denotes a competitive opening and then revealing to your audience that it actually has nothing to do The chart is probably an obvious don't do that kind of thing. So let's look at a more realistic example. So here we have the same visualization, and it's been updated to reflect the renovation period. But there's also the inclusion of a coffee sales line as the manager of the cafe wants the discuss both how the renovation has impacted revenue. Andi, how that coffee has been selling. So these are inherently related, but instantly your fitting the coffee sales into the narrative that you're creating. You can see how an audience will perhaps see this and begin to connect the dots. Despite a renovation period, the coffee sales remain steady. This is probably because most customers buy coffee to go. They don't sit in the cafe because of their presentation. Together, we sought to understand that relationship, even if there wasn't necessarily one there. To begin with your data, visualizations must remain on point on focused. It's in our nature to make connections with the points that stunned out and formed them into some sort of narrative. What you present is crucial to the clarity and impact of the chop. You must ensure that if you present some of them, there's a relationship. Otherwise your audience will instinctively create that relationship, and sometimes they create false narratives to undermine the real information that you're trying to communicate. 11. Visual Perception - Convention: convention convention is the final concept that you should understand. Whilst there may be no universal standard to how a graphics red, you can definitely point out when a graph is just wrong, our brains are wired to see the world in a certain way. Some things that just right and some things are just wrong. Look at this map of the world. Is it wrong? You'd say it's upside down, even though from space there is no right way up. Every man we've ever been shown has been the same way around with America on the left, on Australia on the right. This is lead us to accept that as a standard. So when showing a map that's upside down, it just appears wrong. Even though there's technically nothing wrong about it, certain things are just hard coded into our understanding. Like North is up south is down. Red means stop, halt, bad or hot and green means go or good. Whenever we look at something, we do so through a lens of convention on when things don't conform to it, we just reject them, is correct and you run that risk in data visualization in the world of data visualisation. Some things are just expected. A graph like this one is simply wrong. There's no in factual data in this visualization, but the composition of the graph is incorrect. Titles shouldn't read from top to bottom, lazily hanging out in the top right corner. The legend shouldn't take up so much space placed in the header. There's just the set expected way. Things are to be presented in data visualization. Certain conventions should be followed. The composition of your visualization should be standard. The title, the visualization, the data points, the access, the legend. They should all be in their proper place. Not only should the components of the visualization be arranged in the right order, the visualization itself should also be correct. Here we have a wholesaler showing the number of unit sales to four different stars they've picked. A line graph on the line indicates some linear relationship like time. Where is no such relationship exists between these data points. Therefore, separate categories measuring four separate things. If conventions like this aren't followed, it leads to confusion and misunderstanding. It's a big load. Ian's member may not be able to articulate specifically what is wrong with a child like this, but they won't get a pleasing sense of clarity when viewing. It wouldn't really describe it as insightful. To demonstrate another example, consider this chart that shows product sales from a wholesaler. The similar color segments implied that there's a relationship, but if you look closely, they're a bit muddled up. Bananas. A medium sized T shirts have been grouped by color as well as apples and jeans. The correct grouping would be to group produce and apparel together. If this visualization was presented, the offer would need to fight convention of similar colors, implying relationship. To get the audience to understand the visualization, you should aim to embrace and conformed to convention, not resist all act against it. So we've covered a lot of theory on the basic tenants of how we visually perceive getting these tenants of visual perception right is critical. It's in people's nature to judge information on its perception. If your data visualizations don't highlight the important components, create a simple, accurate narrative, clearly present the ideas UN conformed to convention that it's not just confusing for your audience, they'd actually distrust the information itself. So now join me in the next section of the costs where we will step into the world of visual design and how we can design and bring these visual perception principles, toe life and create impactful, data driven stories. I'll see that. 12. Visual Design and the Application to Data Vizualisation: visual communications are everywhere in advertising websites, book covers, movie posters, illustrations. The list goes on. Visual communications are simply stories communicated to us through a visual medium, such as a picture. For example. Look at this advert. See how without words, it's telling us the viewer a very clear story. The messy house and the loud, screaming Children illustrates a hectic environment, which is seen through the rear view mirror of a bike as the driver leaves the chaotic scene behind and escapes on his motorcycle. Then the picture is punctuated with a single word escape. This is a very solid example of how images can tell a story. A single static image tells us that peace can be found on this motorbike. It appeals to the consumer who wants to leave their typical life behind and escape on an adventure. And this bike offers that all of this was achieved and communicated to us without words with just an image. That's because the designer employed the techniques of visual design. So let's begin by taking a look at the arsenal of options a visual designer has and how these options can be applied to communications. In 1923. Max Wertheimer produced the Gestalt Laws of Perceptual Organization, which outlines some key principles and important ideas for any designer toe. Learn. These principles describe how the human brain will attempt to understand and find meaning in a visual that has many individual parts by looking for patterns that inform the viewer. So these laws essentially describe exactly what is happening in a data communication. Whenever you create some sort of visualization or a graph and show it to an audience, your audience is scanning over that graph, looking at a lot of different shapes and objects and then trying to find order or patterns within them. So let's take a look at these laws and see how they can be applied to data visualization. There are six individual principles associated with the gestalt fairy similarity, continuation, closure, proximity, figure or ground, and finally, symmetry and order. The first principle similarity. So this principle refers to how our minds work group similar objects together. Regardless of their proximity. The objects could be grouped by color, shape or size. Look at this group of objects, weaken group them together automatically without there being any reason to how they've been placed the objects don't even have to be near each other. You can also use this principle to make objects dissimilar to each other. In the world of you. X you I buttons on Web pages are colored or shaped to stand out. This is that principle in action, using color, size or shape to group setting objects or make objects appear different to those around them. When using this principle to communicate with data, you can inform the user of similarity or dissimilarity. We can use this idea in data visualization. For example, take a look at these two charts. The bars across each are are exactly the same book. With the use of color, you can see the principle of similarity in action on the left. The color creates to syriza bars across the three different months. The blue bars related one for each month on the orange bars related also one for each month . Now take a look at the graph on the right. This time there are three groups of related bars, each pack representing a different star. Both grafts used the exact sing bars, both showing a revenue comparison. But because of the relationships created with the coloring. They're both viewed in completely different ways. I can also use the principle of similarity to create dissimilarity in my graphs. In this graph, it shows the relationship between ice cream sales on the temperature. The orange ones have been made to stand out from the rest. They've been created to be dissimilar as they denote when the ice cream was sold out. So those days aren't true recordings for the day, so you can see how big designing around this principle. It makes it very easy for an audience to understand your graph so very quickly. At a glance, your audience can look at this visualization and immediately pick out the three distinctly colored dots. That's because this principle has been applied and these ones have been made dissimilar to the rest. The next principle is continuation. This is the law that when viewing lines, we want to follow the smoothest puff. Regardless of how the lines have been drawn, arise one to follow the straight line from left right from one end to the other. We also want to follow the bend line top to bottom, even though the grey and red lines are separate. This is an important law to remember, especially when you create line graphs. You should be aware of how data could be misrepresented due to the fact that the viewer makes connections that aren't there. A common use of this law in data visualization is to express a forward looking trend or forecast. As you can see in this graph that plots the number of employees monthly, the audience can follow the line forward very easily. If you asked an audience member what the employee count would be in December, they'd very easily be able to tell you that it would be about 118. You should consider this when creating your line graph visualizations because the graph axis had additional space. Even without a forecast line on audience can very easily view of forecast. Andi. In all cases, you may not actually want present forecast. Your audience is This could misrepresent the data, So if you don't intend to show a forecast line, then don't give the visualization the room for the audience to create one. This graph doesn't have a forecast line, but anyone is actually able to tell the forecast, even if I include a forecast line that takes a sharp down with term when audience can still see the upward 10 quite easily. You should be calf of list A. Sometimes you don't always want to show a forecast trend. I also just want to point out that you can see the previous mention law similarity happening in this graph. The Dutch line is clearly distinct from the solid line. Despite making up one line of the same color, the next principle is closure. This principle describes how your brain will fill in the blanks or the missing parts of an image. This law has two farms, a complex and a simple in his complex form. It's how your brain will see the whole image despite missing parts such as this example logo. Despite the missing lines, you can fill in the image and see the panda. The complex foam isn't particularly applicable to data visualization, but in its simple form it can be quite powerful. The simple form of this principle is the idea that our brains will follow a non existent shape such as a lime, which can be used to imply unknown data points, for example, forecasting. So that probably sounds quite familiar to you, especially if you've just come from the previous section. So once again, look at this line graph showing the increase in sales for the last six month by adding the additional months along the axis without having the data points are, brain can fill in the blanks and forecast for the future months so it makes a graph like this, which is seemingly quite simple. On the surface, there's actually an underlying amount of complexity going on in terms of visual design. At first glance, it's, well, just a line, but it's actually a number of different design and visual communication principles acting on it. It's taking advantage of those principles and designing around them. That makes the seemingly innocuous graph really simple and easy to view and importantly, easy for an audience member to understand what it's actually telling you. The next principal, we're going to cover his proximity. So this law describes how our brain will create relationships between objects that are closer together or, in some cases, how the opposite of that is true objects that, if over a part, applying no relationship, for example, these dots as they're all close together, our mind considers them as one object. But if I moved them into rows like this, we now see them as three distinct objects. You can probably tell how this applies to data visualization when you want the audience to compare values or, if there's a relationship between them, ensure their space together. So take a look at this example bar chart. This here is a to Siri's bar chart, which means there are two bars for each category, which in this case means to bars for each month. Because the spacing between each and every bar is equal, the mind doesn't easily see the two bars each month is related. It's like you're looking at 14 separate buzz, even though that colored separately. But if ff in his group together the relationship between each pair of bars is far more obvious, the next principle is figure ground. So this principle describes how there's a very distinct foreground and background and how your brain makes a very clear distinction between them. You've probably seen this famous example by Danish psychologist Edgar Rubin that shows two distinct images of even a candlestick or to people facing each other. This quite effectively demonstrates this concept that your brain interprets his image as having a foreground and background by default. Whatever is in the foreground takes priority and attention in our mind. So this is quite an important concept to be aware of when creating visualisations and communications. If you have too much content in the foreground, it competes retention and focus from your audience, making the message challenging to interpret, typically by default visualizations that you create in tools such as Google Sheets or Excel will place all of the data points on the graph in the foreground, with equal importance, such as this default graph in Excel. This makes it difficult to talk about a couple of bars in particular, and doesn't create a lot of separation between graph elements such as the title and the data values themselves. But by taking advantage of this principle in our design, we can use color to push elements of the graph into the background, thus given more focus to the important parts of the visualization, making it a lot easier for your audience to understand. So by considering this in our design, Cole has been applied that follows this principle. The two day two bars that are important have been brought into the foreground by use of a brighter color on the over bars have been pushed into the background. I've also created more of a separation between the content of the data and the title by adding adult background color to the Axis background. This makes it a lot quicker for your audience to interpret the data. Understand the visualization. There's not so much in the foreground competing for their attention. There's a very separate foreground and background where the important components of the graph for in the foreground and the less important ones to the story have been pushed into the background. The final principle that we're going to touch on is symmetry and order. The law of symmetry, say's that your brain will perceive an image in the most simple way possible. For example, have you ever seen a shape in the clouds or recognized a face in the shape of a bush? This is the principle of symmetry in action. Your brain interprets any visual image in the most simple way possible. For example, the shape on the left is in the complex. Enter gone, a nine sided cheap. You actually perceive it as a triangle, a circle and a square if you call back to an earlier section of the course where we discussed how our brains can't take in too much information and it will find trends and data quite easily. Well, this lower visual design complements the idea in the world of data visualisation referring back to this example. Visually, you perceive as one shape and not the 15,000 shapes that actually is. This is the law of symmetry and order in action. It allows us to see a large data set and easily pick our trend within the data. But if you actually wanted a highlight a few examples at this 15,000 then you're probably gonna have to rely on one of the over mentioned principles earlier. 13. Components of a Data Vizualisation: before we proceed with what the different types of graphs are on which ones are most effective. Let's first learn some common terminology on all the components that come together to form a data visualisation. So here we have a very typical data graph. It's showing you the revenue for four regions for both this year and last year. So let's get familiar with the different components of this visualization, so that when you're creating your own, you know what each and every component is that you can refine to tell your story. In the preceding chapter, we'll learn how each of these components can be designed to create an impactful data communication. So firstly, we have the title. This is a short description used to inform the viewer of what the visualization is all about. You might recall familiar that the title isn't normally the first component seen by the audience, but it does give them an understanding about the visualization. Next, the title can be supported with a subtitle that offers more clarity on what the visualization is all about. Finally, the sauce. You'll find it typically hanging out in the bottom left of the graph, and it informs the curious viewer of where the date was sourced from. For internal data visualization, it's not always necessary, but is best practice toe. Let the viewer know where you, as the author got the data used in the visualization, so that's what you'll find around the typical data graph. So let's inspect the actual visualization itself. The actual graphical representation of the data in this case is bars, but in over data graphs, they could also be segments, bubbles, lines or points. In this example, we have two different Siris of bars, which represent each metric. Some, but not all, data visualizations will have multiple dimensions or Siri's that they display, but we'll get into more of that later. Now each of the bars is accompanied with a data label, which is usually the numerical value that the bar off object represents. Most common visualizations haven't X or Y axis, which is accompanied with, but not in all cases, axis labels, access titles and access lines. And finally, we have the legend, which informs the viewer of what Siris of bars are over objects, represents which of the metrics being displayed. So that's everything that goes into a data visualisation. All of these individual components come together to visualize your data and story. Each of these components must be considered and incorporated into your design to tell an effective narrative. These are the components that form the most common and standard data graphs. But there's many more types of data visualisation out there. So now join me in the next video where we will take a lock up both the common andan common data visualizations learn what not to do with data visualization and then proceed into picking the right graph for the right data. 14. Different Types of Graph: so most people will be aware of the major types of data visualisation, such as the line chart or the bar chart, but is actually many more available to choose from now. There's no universally agreed upon way to categorise all the different graphs. But any analyst would agree with the following categories comparison time, Siri's ranking correlation, geographical and then a few that I don't think fit into any of these categories I've grouped under other on Don't worry, you don't have to remember everything about each of the visualizations that I'm going to be explaining, as at the end of this I'll show of you a job aid that helps you pick the best chart for your data. But do familiarize yourself with the different options when they're applicable, and particularly when they aren't the first category visualization that we're going to take a look up is comparative. The comparative visualizations are best used when you wish to show a Siris of values to be compared. They're very effective at showing one or more Siris of values so a viewer can very quickly see when a certain value is bigger, smaller or comparable to some others. The comparative graphs are essentially variants of the body shots. The Bata, the bread and butter staple of data visualisation attributed to William Playfair in his book The Commercial and Political Apus, first published in 17 81. As a self confessed data nerd, it really does excite me to show you the first ever use of the bar chart. Now this visualization could benefit from a few pointers from this cost to make it more impactful. But nonetheless, it's effective. The reason the bar chart has stuck around for so long is because it is just so effective and versatile. As we proceed for the different categories of data visualisation. You'll notice that our friend the Bar chart makes a few appearances. Onda, as you well know, visual understanding. It's a lot about how our brain can perceive on our brains are exceptionally good at gauging the length of a buyer when it's been encoded as a value, which is essentially, what about you actually does. We can instantly see when a bar is bigger or smaller, even if its just by a little bit, and we can easily understand what value that little bit represents. We're also very good, accurately gauging the value of bar represents when compared with poses to demonstrate. Look at this graph. The bottom bar is 100 on at a glance. You can see the second to last bar is 75. The second bar is 50 on the first bar is just below 50. At 45 you can very quickly and easily interpret those values without them needing to be pointed out on the graph. You can do so visually because our brains are just so good engaging the length of a bar. The comparative charts fall into two categories. The comparison between values and the comparison to the whole. The comparison between values Siris of charts consists of the body shop, sometimes known as the column child. If the bars of vertical, this is the standard way of showing comparative data, the horizontal bar char or simply bar chart. If the latter was known as the column chart can be useful, particularly when the bottom axis isn't time or you have a particularly long access label. You also have the group bar chart variants of these if you're showing multiple Siri's so the bar is the most common shaped to use in this category of charts. I mean, it is called the bar chart, after all, but you may have come across bar charts that use various different shapes, such as in an infographic. I'm gonna recommend that generally you avoid using charts like that. Think back to what visual perception theory of Convention said. The bar is clear, understood on the common gnome. So stick to what you audience knows we'll be covering using different shapes and data visualization in the next section. The deadly sins of graphs. So next we have the comparison toe hole when you want to show you audience how something well compares to the whole these air the graphs best suited for that job. For example, suppose you have numerous survey results, and you want to display an age breakdown of all the different respondents in the comparisons. A whole category of charts you can choose between the stacked column, which is quite effective and could be used to also show over time if you use the bottom axis as a time Siri's. But there are a few rules that you should follow If you want to make sure that this graph is effective. Firstly, you should ensure that no segment is too small to be seen under. You also shouldn't show that you don't have too many segments if you find that you have need to display such data that I would recommend picking a tree map as it's more effective . The tree map is also sometimes called a grid map or an area map. In fact, you'll find that a lot of graphs go by different names, particularly between different visualization tools. A tree map is a series of rectangles where the size of each rectangle represents a value with a tree map. You can also add a lot of data density. That's because you can use color to represent another value. You can even introduce sub categories by grouping different rectangles under a heading segment. The next category of chart we're gonna look up is the time. Siri's, a particularly common visualization is one that shows change over time. That time could be anything between seconds, hours, months, years, decades. You get the point. If you need to communicate how something has changed over a period of time, then I would recommend picking between these graphs. First up the line graph, possibly the most common use of showing data over a time period, but do be careful not to plot too many lines. Otherwise, you end up with its ineffective cousin. The spaghetti chart. Having too many crisscrossing lines makes it quite challenging for an audience to read and understand. I mean, just look at this thing. I have no idea what's going on here. I don't think there's any amount of annotation that could help explain this. So next up we have the area chart. It's just like a line chart, except it's been colored in. It just shows the total change over time. One of the key principles in designing to communicate is to maximize the white space so that you can draw a clear distinction between the foreground and the background. This particular job goes against that by using a lot of cold space. I mean, it really is just a line graph except with a less effective design, so wouldn't really recommend using this one personally. But having said that, there is a police for its use. Remember the visual design principle earlier about how our minds want to continue following a line, even if it takes a turn to remind you his a graph. The solid line and the dash line are each representing a different value. Our minds want them to cross, even though they never do. Even if I show both lines is quite separate by making one a dash lime, it's still a bit confusing. Now it is possible to understand this graph, but you're fighting your natural propensity to follow the lines as they cross. So if you find that your graph makes it hard for your audience to follow the lines, then perhaps the area graph is the more effective choice of line graph. As it makes it easier for the audience to follow the line correctly, you can see how in this example, using the area graph makes it easy to distinguish where the lines don't cross. Next we have the bar or column graph. I love the bar graph. It's so versatile that if you put the bottom actors this time, then you can use a bar graph but similar to the line graph. If you have a lot of Siri's to show that it's not a particularly effective choice, if you do find that you have a lot of Siri's to show over time, then a slow graph is probably the best option, but it does limit you to before and after picture, as opposed to multiple points over time. So the next category is ranking of data. Once again, our friend the bar chart makes an appearance by using a horizontal bar job. You can display your data in order of ranking. If you want to show how ranks of data have changed over time, then a slope graph is also a very effective way of doing that. So next up, we have correlation. If you want to show the relationship between two continuous variables the new console ECT, the scatter plot. Andi. If you've got three variables, the new comm picked scatter plots, cousin the bubble sharp, where the sides of the circle is the third variable. So finally become to the geographical charts. Geographic regions such as the country's or state's could be colored or used of icons to represent data from those geographic regions. There are, however, a few considerations when using this type of graph, we'll get more into them in the next chapter. But personally, I don't recommend using these graphs when communicating with data. They're quite flashy and visually appealing, which leads people toe want to use them, but they have a few downfalls and actually not particularly effective, representing data accurately but more on that in the next section, so that just about covers the different categories of charts available for you to select from. But there are a few more charts that could be used effectively that don't really fit well into any of these categories. So let's talk about what over graph and data visualization options you have available. Firstly, if you want to measure to a target, then you can use the simple gauge a very simple and elegant way to communicate how far a single measure is to a target of first glance, the gauge looks quite simple, but there is quite a lot of information contained in this visualization. You have the current value, a starting point, an ending point, a target, and you can even color the bar in to show an additional metric. So quite a lot of information is packed into a small space, which makes the gauge quite effective at telling a specific story. It does, however, fall down when you're trying to make comparisons with values or giving an overtime view. It can't show that over time. Change on. They do take quite a lot of space or what they show, so you can't really have too many gauges on one page. A table should definitely not be overlooked When it comes to effective data visualization, a well designed full month, a table is a pretty good choice in a lot of cases. Now you have two different possibilities where a table is probably your best choice. Suppose firstly, that you have a lot of data points to display on your interested in communicating the outliers. Take, for example, this scenario. The head office of 15 different call centers wants to visualize the average survey results , which are on a scale of 1 to 10. They want to view these monthly results over a period of a year and see which mumps were outliers. A visual ization like this table could be used in this case, using formatting and a color scale. The worst month very quickly stuns out to your audience, but it also displays that month in the context of the whole year, and it allows for a few outliers like start to in April to be discussed. So you got 15 different stars over 12 different months. There's a lot of data to consume, and this table does so in a very elegant way and allows the audience to pick up the story, the outliers and see the context for the whole year as well as being able to drill into any particular month or any particular star. So you might be thinking that Well, since we're looking a data Siri's plotted over time, then why not use a line graph? So this is the same data in a line graph on what we've ended up creating is just a very messy spaghetti child. You can drill into any individual store. You can't really effectively see the average survey results over any month. Sure, there is the big dip that stands out, but you certainly lose all the context and detail that the table so elegantly puts across to the audience. Okay, so let's clean up this line chart. Perhaps if we just average out the server results for each month, that will make things easier to read Now. This does show quickly which month was an overall outlier but it doesn't allow for discussion of the other out liars that the table allowed far. Or can anyone discuss the individual call centers? Sometimes a neat for matter table just works best, particularly if you employ the use of color. The second great way a table can be used is if you've got to communicate the audience lots of varying pieces of data about a particular Siri's. So you've got to do a comparison. You need to show things over time. You've got to show things ranked to the whole. You've got a lot of different data that just doesn't really fit on any particular graph. Take, for example, this visualization that shows the marketing Dichter from a series of surveys sent out for an ice cream store. The visualization contains seven Siris of data, one for each age range group, and it shows it broken down by the male female split across three entirely different metrics, the percentage of respondents the favorite flavor on how much they spent install when combined with a bar graph. This table is a very effective way for an audience to consume and understand a lot of data across a lot of different metrics. You would have really used this type of visualization if you have a very poignant story to tell. Such as the females have the least variance in favorite flavor. But a visualization like this one is a great scene setter to provide the context. If you then want to go on to communicate specifics about the marketing result on the very last visualization that I want to touch on is perhaps the most complicated. Just the number. Yep, that's right. Sometimes don't overestimate the power that a simple number could bring. Sometimes you don't need to overthink your visualizations, as the number can often reveal more than you think. If you have an audience that has all the necessary context sometimes and number can be the most effective way to tell a story. Take this scenario is an example. Suppose you work at an organization that has recently rolled out a new health and safety compliance training, and you want to ensure that everyone in the organization has taken this training. You've been asked to update senior management on the progress of this compliance training. They want to know what the level of complaints is on, what the action plan is to ensure that noncompliant individuals take the training. So you take a look at the numbers and you see that over the last few weeks, the number of people who have taken the training reaches 99% compliance, which is a great success that you want to share with them. But you have over 1000 different members of staff to record. So how can you best communicate this of impact without getting lost in the detail? So since you've got data over time, you therefore consider the line graph. Well, it does show how the compliant has changed over time, but doesn't really emphasize the fact that compliance is so high. It doesn't really show the success of the training. So perhaps that since it's a single metric, you show it against the target. By using a gauge, which is pretty good. We don't really feel that it drives home the result with as much impact as it could. So in the end you decided to just let the numbers speak for itself, which in a lot of cases is all you need to do to tell a great story when telling great stories of data. It's not about a really complicated graph. It's more often about just telling the right number to the right audience at the right time . You think back to what we learn from Friedman. Often, time design is create really complicated, beautiful looking graphs that ultimately say nothing. A well designed and formatted graph doesn't inherently say anything. It's the number that does the speaking. So sometimes you only have to let the numbers speak for itself. In this small case study, we explored how the number was presented in different ways, but none of them really drove home any impact of the audience because we wanted to sing the success of the 99% compliance. So just putting 99% compliance is the best way to do that. You don't need a complicated graph in all cases. 15. Deadly Sins of Graph Design: however, when creating impactful data visualizations that will resonate with an audience. There's a lot of considerations you have to make on what you should do. But there's also plenty to consider on what you shouldn't do. As a data visualization expert, You should be aware of the ways in which data could be used to mislead or bring your audience to false conclusions so that you can avoid making the design choices that lead to misrepresentative graph. So join me in this section of the course where we will cover the deadly sins of graph design. It's easy to lie with statistics. It is hard to tell the truth without it. So just reflect on that for a brief moment. What does it mean to say that you can lie with numbers or tell the truth without them? Surely the number is the truth. Well, not necessarily. And certainly in my career I've encountered a number of people who are immediately quite dismissive of data and graphs and visualize a zham. They say things like, Oh, I don't want to see the numbers. I want to see the truth and I take a step back and think, well, the numbers are the truth, aren't they? So in this section, we're going to explore how you can torture and twist numbers to know inherently lie. But they're certainly not always telling the truth, at least in a representative accurate way. So in statistics, there's a term misleading graph or sometimes distorted graph. This is a graph that, through intentional effort to mislead or accidental poor design, is a graph that just misrepresents the data and encourages of false or inaccurate conclusion. I've certainly seen a lot of graphs in my time, especially in the workplace. On on more than one occasion, I've seen graphs that have been created accidentally to misrepresent the data. That's because the offer just didn't know what they actually did was quite misrepresentative. They told the incorrect story due to the poor design choices that they made. And aside from accidental unfortunately, graphs like these are done deliberately all the time. You have probably encountered on many occasions thes types of graphs in news media. Suddenly in particular, when it comes to politics, this is a strategy that they always take reported on the news all the time by different politicians for different political parties are graphs there intentionally designed to misrepresent the numbers to try and influence the audience. In one way that isn't truthful. For example, take a look at this graph. It's known as a truncated graph. That means that the Y axis doesn't begin at zero. In this example, this was done to make the difference between the yes and the no vote seemed far more extreme. That actually waas compared to this graph that shows the same data without the Y axis being truncated suddenly, that extreme difference in the leave and remain votes doesn't seem so extreme anymore. You can imagine how viewing this graph my influence an audience. There's nothing inaccurate necessarily about the data. But the way you have presented was certainly misrepresentative. That's what it means to lie with numbers, the numbers themselves on incorrect. But the certainly lying to you when they tell you a story. So I'm confident that you're an honest person who wouldn't go around and mislead with data intentionally. But sometimes a misleading graph is created due to its poor design. Visualisation tools such as Excel will automatically truncate the Y axis. If many of the data points plotted close together, so we're going to learn the ways of misrepresenting data. Not so the you can go out there mislead but to quip you with the knowledge to spot when you're being misled and to avoid accidentally doing so with your design choices. The ways of creating a misleading graph could be categorized in two ways. Firstly, the shape of the graph and then, secondly, the layout of the elements. So let's dig into each one of these categories. Discover why they create ineffective and misleading graphs and learn how we can design around them. Firstly, let's explore the graph shape and see how it can be used to mislead your audience and misrepresent your data. So picking the correct shape for your visualization is critical to getting your message across We covered earlier. Which grafts best represent the type of data you have on the message that you want to get to your audience. But these are the graphs you should definitely avoid using. Firstly, the three d graph the three D graph is something that should be avoided at all costs. It's all far too easy to put your data into a graph visualization tool such as Excel or Google sheets and then pick a nice bar chart to represent that data and then look through the formatting options you've got available. And then you hover over the three D graph because looks kind of appealing. Well, you must never click that button. You must avoid using three D graphs. Three D graphs are a blight on the world of data visualisation all to after nine counter them in the workplace. There seems to be a common held belief that the three dimensions add some sort of authority to the graph. Data is inherently a complicated thing. Therefore, adding more complexity to the visualization must make it better, right? Well, no. Adding three D introduces graph elements that only add to the clutter of your graph on your audience finds thes distracting. Remember, we want to maximize white space in our graphs, not detract from it unnecessarily. Also, three D graphs introduce 1/3 dimension or metric, but isn't actually measuring anything. The way the bar extends backwards isn't actually encoded with a value. So we're showing a shape which represents a value to the audience. But we're saying the only one of those dimensions actually has a value. It doesn't matter how far back the bar extends. It doesn't actually mean anything. So what we've done is really clutter up the visualization, but not actually introduced any metrics that can be measured. So those air to kind of softball reasons to avoid three D graphs. But the number one reason you should avoid using three D charts is that the third dimension means it's really easy for your audience to misread the graph, adding that third dimension allows a lot of ambiguity to the graph. It makes it quite difficult to read and introduces a lot of room for misreading an error. So what I want you to do is take a look at this graph as an example of how the three dimensions can make it challenging for an audience that accurately read it, this graph is showing you revenue for six different stores. What I want you to do is look at Star fall and answer the question. What was the revenue generated? Pause the video, if you need to, and then remember the answer that you gave So the answer? Well, it's not exactly clear. I think it's around 5000. The problem is how exactly do you measure a three d bar? Is the bomb measured from the front face like this? Okay, so let's measure it from the front. In this case, the bar represents 4500. Well, maybe that's incorrect. What if we measure it from the back? So let's take a measure from the back where the bar touches the invisible plane of the wall . Now it measures fall 1009 100. Well, what if we follow the line leading us back to the axis? If we do that, it measures. And even 5000 Well, in fact, the bar represents 5149 so visually, just looking at one of the six bars you can easily interpret the bar is being between 509,000 which both of these still lower than the actual amount. That's an inaccuracy range of over 20% and that's just one bar. Imagine the confusing the audience could experience trying to interpret how all six of these bars compared to one another and the amount of inaccuracy that could come about from that. The problem is, there's just no stunned ID way to measure a three D bar picture a scenario where you have three different people viewing a graph like this, each measuring all six bars in a different way from one another. Ah, lot of confusion is going to be had. Now take a look at how easy it is to see the value stall fall when the bar is to D, you can very quickly and easily see that the bars around 5100 using three D is not just a personal great. It's one of the golden rules, so just never use them. Suddenly, three D graphs are commonplace tactic used by those trying to mislead you with data. So be on the lookout for this, unfortunately, partisan and politicized issues often represented with three D graphs by those trying to push a certain agenda or viewpoint. Often they use even more extreme view angles or stretching the third dimension fervor into the background, which only serves to make it harder for you to understand the graph. So hopefully you can now spot when this tactic is being used on you and avoid viewing that graph or misinterpreting it. So isn't just three D shapes than a misrepresenting its most over shapes. In fact, if you call from a previous section, we touched on why the bar is so effective to remind you it's because our brain can very easily and accurately gauge and compare the size of bars or rectangles off. The shapes were not really very good up to demonstrate. Take a look. How these shapes scale each of these three shapes is very misleading when it comes to scaling them. See how the size of the square is only three times the size in scale, but it's already nine times the size in volume. Now look at the triangle in the circle, the increase in size very quickly and surprisingly. And now look at how the bar scales up again. Very easy to quickly gauge. Compare the size of a bar. These are the shapes to make it very difficult for an audience to properly get a sense of the size of those shapes, which is essential if you want your audience to properly compare and understand them. Take, for example, this triangle graph I want to ask you how much bigger is the second triangle compared to the first? Well, it's probably about twice as big Mm. Actually, it's only 50% bigger. Now that I've said that, you're probably saying, Ah, yeah, I can see that now it is 50% bigger. Well, I lied. It's actually only 1/3 bigger. But see how much room for error there is You or over? Audience members could probably believe that it is twice as big. You'll certainly believe that it is 50% bigger when Niva true. So now look at those same values Baugh's. There's just no way. It's twice as big or even 50% bigger. It's plainly obvious that the second bar is only 1/3 bigger pitch is used in grafs aren't too commonplace when it comes to data visualization within a company. But sadly, when it comes to public data visualization, it is a common practice to use such graphs to mislead you. So you might be thinking, Why would I ever use a triangle chart? It isn't even one of the shapes available in graft, creating tools such as Google sheets, which is true, but you may have seen shapes in the form of pictorials in infographics. It's quite common in infographics to use pictures or shapes to represent the data. But the same principles apply. See how the principles of shape scaling apply to these pictorial graph examples. The man between A and B is three times bigger. Book is nine times in volume, which is quite misleading and confusing for an audience to properly gauge the size. You create similar problems like this for your audience when you use pictorials to represent single values, such as in this chart as the banana, apple and cherry, or all different sizes but each used to represent a single unit, they create different limbs. The length of the rover Cherries is half the length for bananas, but both represent the value fall. When it comes to data visualization, just stick to the common shapes and avoid pectorals like these. Take, for example, this gruff that shows you the revenue of different food outlets. These images have been scaled by height book that two D images, meaning as you scale by the height, you also scale by with. So as we increase the size, we actually double that size in volume, giving the impression of far greater growth. Take a look at how this graph exaggerates the size of the McDonald's logo it's actually less than four times the height of Burger King, but it appears to be about 12 times the size. This is the issue of scaling images. They exaggerate. So be on the lookout for this in news media and online. When you see a picture graph like this one, the author is probably trying to mislead you on the topic of misleading shapes. I think it's about time we addressed the pie chart. I'm giving them their own dedicated section because they're just so bad and prolific that they need their own dedicated moment to document just how much I and any other data visualization expert hates pie charts. I'm gonna come straight out and advise you to never use a pie job. It pleases me to go on record saying that the pie chop is the worst chart to possibly use. Yet somehow it has such widespread use in the workplace. The pie chart is such an ineffective chart, and yet somehow, despite that, it has such prolific use in the workplace. In fact, I'm sure at some point you've probably been guilty of using a pie chart, so hopefully as we go through this section, I can convince you to never use one again. So I want you to promise me after this costs your bun pie charts of any future reports off visualizations that you create. So what exactly makes the pie charts so ineffective? Why is it the scourge of data visualisation? Well, pie charts are just ineffective at comparing values. They're very distorted, and they make it extremely hard for decision makers. Toe understand the message that they contain. There's two key issues with using a pie chart, firstly, that misleading, and secondly, they have a slew of limitations. So let's explore what each of these mean and why they make the pie chart so terrible. Firstly, much like the three D shapes that encounter scaling issues that we learned in the previous section pie segments a hard for us to measure and compare accurately, which is, after all, the main point of data visualisation. I want you to take a look at this pie chart and answer the question. How much bigger is Segment two compared to Segment five? So, firstly, I bet it was a little difficult to work out which segments two and five actually are, which is just a number of the problems of pie charts on the face of them. They're just challenging to understand because they're layout requires you to look back and forth between different elements of the graph. No graphs should have their audience darting between all the elements, just trying to figure out which segment represents what, anyway, Back to the size comparison. It appears as though segment to is about 20% bigger, but this is far from accurate. Section two is actually 150% the size of Section five, so you can see how the pie chart is shaping up to be pretty ineffective on audience member viewing. The pie chart could at first just need to work out what segment means what and then they will start making inaccurate comparisons. But that's not all. The misleading size comparison is compounded if you make the biggest sin in the world of data visualisation the three D pie chart. So here we have the same chart, but this time it's committed the mortal sin of becoming three D. Now, Section two looks twice or even 2.5 times the size of Section five. These are actually the same graphs. But now the segments look vastly different when compared. So now compare section three to Section five. Which one is bigger? They look about the same. Or perhaps Section three is a little bigger. I'm sure you know where this is going. Section three is actually a lot smaller, so you can see the problems of representing day. To this way, it's impossible to measure accurately with visuals, So you could be saying, Well, let's make it easier on the audience by labeling the pie job so that an audience can understand which segments represent what? Which leads me to sort of the next problem with pie charts the limitations they have to make this pie graph understandable. I've had to include labels for the legend. The pie segment name the value for each segment on the percentage of each segment. If you're going to write everything down, then you're eliminating the point of data visualisation, which is to make data quickly and easily understandable. The next limitation of pie charts is just how quickly they get messy. In this example, we only had five segments on things are already beginning to look crowded. In a best case scenario, the pie chart is a limited to a single Siris of data that has a maximum of five values. Beyond that, the so many labels and elements to the graph, the audiences mind begins to get overloaded. And then they stop reading your message, which is something you definitely want to avoid. You're supposed to be making the lives of the decision makers in your audience easier by visualizing the data. The pie chart is just so ineffective at doing that. Not only is it inaccurate, it's also a poor design. You need to label everything, write everything down, which, if you're going to do so, you may as well have just written it down in the first place, which is a simpler and easier way than trying to read everything on a pie chart. So a pie chart fundamentally doesn't fulfill the purpose of data visualisation. So one of the next issue two pie charts. Not that we really need another issue by now hopefully have convinced you to abundant the pie chart. But another issue they have is they take up a lot of space on the page. Space on the page is often a limitation when doing a presentation or apart. You have a finite amount of space on the page to get your message across on a pie chart takes up a lot of that space unnecessarily. Concisely, communicating and maximizing space on the page is a key part of an effective data story. So why pick the pie chart that will take up more space to get the same data across than another option, such as a barter up? Look at this side by side comparison of the same data and know how much more space the pie chart takes up to display the same data as the bar chart. And it's doing so less effectively when space is such a constraint to communicate a message , you can't afford to waste it on by using a pie chart. It's utilizing precious space unnecessarily. So one of the over design issues of the pie chart is the fact that there's just no white space. It's all color, all of it competing for the audience's attention. We know from an early segment that we want to maximize the white space in our graphs so that we can create a clear distinction between foreground and the background because the pie chart is just all bright colors and no background. Everything is placed in the foreground of the image, all of the different segments of competing for the attention of the viewer. When that happens, it's like three people trying to walk through a doorway at once and just getting stuck so that none of them could get through. When everything is trying to grab your attention, none of it ends up actually grabbing any attention. When all of the pie chart segments all brightly colored, the audience cannot really absorb any off them in any effective way. Okay, you could try and create the distinction between the foreground and the background by coloring the pie segments with muted colors and highlighting the important ones so that they stand out. But the only way to really identify the different pie segments is with the use of color. So if you call the pie segments the same, that now you end up with is just a circle right. Here's an example. Have this pie chart, and I've tried to create the foreground in the background by highlighting the segment that is important. But now how you're supposed to tell which segment means what? So now that you've done this, you're forced to label everything, which again, if you're going to write it all down, then you might as well have just typed up a report, since it's easier for the audience to read. So now look at the same graph is a bar chart the highlight? The difference of how easy it is to create the foreground in the background of my data. Elements on highlight the ones that are important, all without losing the ability to read the chart or needing to explain heavily to the audience Just how this thing works again. Just don't use pie charts. And so we've reached the final problem with pie charts. By now, I hope I've done a good job in convincing you to never need one. Making this final problem kind of redundant. But if you still need a little bit of convincing than the final issue with pie charts is that they're always a circle, and when things have always the same solid shape, it implies that there's kind of a static hole there comparing to when there isn't necessarily one, which makes comparisons between pie charts impossible. Suppose thes pie charts represent the book sales of two different months. When you look at these two grass, the only comparison that you can make is what percentage of sales each book made compared to the other books of the same month. You can't compare between months. For example, you can see that in March, Book one was about 1/3 of the total sales for that month and in February. Book one was also about 1/3 of the sales for that month. But what you can compare is a total sales these books stars made between the mumps. Okay, so we know the book. One had 1/3 of the sales. But how many sales is that? And how many total sales are there within the month? You can't tell between pie charts because they can't be compared. The pie chart is a circle, so no matter how many or few sales happening months, it remains the same size shape. Look at how the data changes when represented. As a Bachop, you can clearly see that march was a farmer successful sales month, which doesn't come across a toll in the pie chart. So, in conclusion, never use a pie chart. I have a simple rule that you can follow pie donor and spaghetti, their food to be eaten, not charts to be analyzed. So far, we learn how the graph themselves can use different shapes to misrepresent and distort the data on ultimately mislead and audience. In this section, we're going to take a look at the graph elements themselves and what you can do to a good graph, for example, a Biograph and how you can then twist it into misrepresenting data and misleading an audience. So we're gonna take a look. Truncated graphs. If you remember familiar, we briefly touched apartness to remind you a truncated graph is when the Y axis doesn't start. Zero. Say, for example, a graph that goes up in increments of 10. The Y axis might start at 50 instead of zero, and then continue on from 50 to 60 70 80 and so on. When you do this with the bar graph or line graph, it over exaggerates when there's a trend to demonstrate. Take a look at this graph I have here. It clearly shows a very strong upward trend, but what you might not have noticed is that the Y axis has been truncated, which is far, in a way, exaggerating the trend more than actually is. If you look at the same graph here, which shows the Y axis beginning at zero. Suddenly, the upward trend doesn't appear to be so strong. Fact, it's barely noticeable. So this is why the use of truncated graphs is heavily discouraged. They overly exaggerate when there's an upward or downward trend. You should be aware that some data visualization tools such as Excel will automatically truncate the Y axis in the default graph. If the values fall within a narrow range, you have to then manually go back and change it so the Axis starts from zero. So just be aware of that when using such tools. So it is sometimes tempting to truncate the Y axis. Perhaps you're trying to save space on the visualization, so you truncate the access and then you add some kind of marker toe. Let the audience know that it's been truncated. Well, you shouldn't do that. Even that is discouraged. Remember, data visualizations are visual representations of the data. There's no point showing the data visualization that is inaccurate and then labelling it and telling the audience that this visual representation off the data isn't actually a visual representation, which is essentially what you're doing when you truncate a why access, even if you label that has been done, studies have shown that even when the audience knows that the Y axis has been truncated, they were still substantially overestimate the actual values. So even if you displayed a graph such as this one and indicated that the Y axis had been truncated, the audience was still see the upward trend as far more extreme. That actually is so moving on to another way to misrepresent data is to play around with the Axis scale. It isn't just truncating the Y axis that can misrepresent data. It's changing the scale of the Y, or even the X axis that can completely misrepresent the data to your audience. Changing the scale means to change the minimum and maximum values that a pair I want you to take a look at these three graphs. Now the data between them is exactly the same. But the scale of the Y axis has been changed between each of the three. You can see in this example that each of the three line graphs shows the exact same data. But because the Y axis scale has been changed, the trend seems either extremely high or very shallow. You get a simile misrepresented result when you change the dimensions of the graph again. All three of these graphs represent the exact same data. But the dimensions of the space the line has been plotted on has been adjusted to manipulate how exaggerated the incline of the line is. So another way to exaggerate or minimize the differences. It's a show, a graph with no scale. The lack of a starting value for the Y axis makes it unclear whether the graph is truncated or not. Additionally, the lack of tick marks prevents the reader from determining whether the graph bars are properly scaled. Without a scale, the visual difference between the bars can very easily be manipulated, So the final way to misrepresent data is to do so geographically. Geographic maps aren't very good representing date accurately. They look quite flashy and aesthetically pleasing, but they do have a few issues when representing data using a map like this one, where the space of each area, in this case the countries has been colored on a red to green scale, indicating the sales of one month compared to the previous month, meaning if sales were above the sales of last month, it's green. And if it was below its red, and if it's similar, it's been colored orange. Now this isn't a very effective way to represent this data. That's because different countries are different sizes and have different populations. We instantly give more focus and attention toe largest spaces on the map, which doesn't always correlate to the importance of each country. For example, in this case, Canada takes a lot of focus, and it's green showing that sales were above that off last month. But this particular company, in this example, has a very small presence in Canada, so that huge green area actually represents very few actual sales. Compare that to the UK, which has a very small area which represents 60% of this imaginary company sales. Well, that huge difference doesn't come across on the map. It makes it a pair is countries such as Canada and Russia take huge importance, and it downplays the importance of the UK. So that's the issue of using a sort of geographic map to represent data. They do a very poor job in giving the context of the audience. You can't compare anything between a map. They're actually very two dimensional. They don't have a lot of data density yet they look kind of flashy and appealing because they have. Ah, they're a cool way to represent geographic data, but ultimately they don't do a very good job of it. But that isn't to say you can't use a map. If you are going to use a map, then you should use an indicator on each country. Are not color in the boundaries of a country. Let's recap the sins off graph design so you can avoid thes mistakes when telling your data stories. So there's two ways to misrepresent the shape of the data on manipulating the graph elements. When it comes to the shape of data, avoid infographics, pick topples and any off a shape over than the lines on the line graph and the bars on a bar chart. That's because, as humans were exceptionally good at gauging the size of a bar, but because any over shaped scales in two dimensions were very bad, engaging them also, when it comes to shape, avoid any three D charts altogether. They don't add anything to the visualization that make it far more confusing for an audience and introduce a huge margin of error when your audience reads the values on the graph. Finally, on the subject of shapes. Never use a pie chart. They're inaccurate, very limited on din. Apparently poor visual design. Any time you can use a pie chart, just why pick it when a bar chart does the job? A lot better. So the second category off misrepresenting data is to manipulate the graph itself. This is done by using a truncated Y axis, meaning it doesn't start from zero. You can also achieve this misrepresentative result by just ing scaling, removing the scaling altogether or just changing the dimensions of the graph so that it distorts the data. So now join me in the next section, where we will see these principles of misleading graphs used on riel life examples that have been published with the intention off misleading. I'm a bit of a nerd if you probably couldn't already tell on over the time I've collected some terrible or misleading graphs, so I invite you to take a toll of me through my data hall of shame 16. How to Avoid being Mislead with Data: hi there. As you well know, Riel life, examples of data visualisation, Not everywhere you look. I bet that have, in the last few days you've seen some kind of published graph in the news online shared on social media or published is part of a study. Unfortunately, we live in a time where the intentions of the authors of these graphs are to mislead, admits represent. Suddenly, this is a very popular tool of politicized issues such as gun control, global warming, uneven election votes, different parties or media outlets that favor a particular party. Were published graphs intentionally mislead the audience to view their party favorably or the over party unfavorably. Either way, they aren't truthful. Intentionally misleading graphs is also a tactic used by companies to misrepresent their activities. For example, fizzy soda companies showing studies that against better logic seemed to downplay the negative results of what they actually produce. These types of graphs are everywhere. Often you might see a well made, at least formatting wise graph online in perhaps the news Onda assume. Since the news is an authority on the subject and it's cold, hard datum that they're presenting, then you think OK, this is fine, but if you do so, you're at risk of taking the numbers at face value without considering the truth behind the matter. So in this section of the costs I'd like to review, some real graphs have been published online in an attempt to mislead their audience. I'll break down how these graphs are misleading and then hopefully a quick new with the ability to second grass graphs and avoid being misled on matters. So let's take a look at my whole of shame. So here we have the first graph that takes pride in place in my hall of shame, of terrible graphs of attempting to mislead the audience. So if you want to pause the video and just get familiar with the graph or take a look at it in the book conned out, you could do so. But let's start digesting this graph. What is it telling us? Okay, gun deaths in Florida, right? So we're looking at the number of murders committed using firearms in Florida along the Bund, and we have a bit of a timeline from the early nineties open till 2014 when this graph is published on we have the number of gun deaths a year on year from the nineties to 2014. And then we've got this label on the graph 2005 floater enacted. Its stand your ground law. No, I'm not gonna go into detail about that low, but essentially, you can use deadly force when your life is in danger, which means shooting people. I guess I'm not gonna go into the details of that. Let's take a look at the graph. So we seem to be implying that there was a high number of deaths and then this low is enacted on the number of gun deaths plummeted. So it looks favorably on this law. Or does it? Is that what the hatch Shell data is telling us now? I think you may have noticed the terrible design misleading design choice in this graph. If you haven't take a look at the Y axis, the Y axis starts at 800 on, then goes up to zero. That's right. This graph is upside down. So whoever published this graph whoever see Chan, who the name of the author is written on the graph. Whoever this person was, they attempted to mislead you with this graph. They look favorably on gun laws. They pass, published all of the number of gun deaths and then saw that it told the opposite to what they want to present. So they simply foot the graph upside down and go he go the this is the data. So this is just, you know, I recommend to anyone don't don't do this, but the reason we can fall for graph like this, or even though that there's nothing incorrect about this data a little bit, I'll get into that. But it's nothing incorrect upfront about it, right? It's showing you different data points on its labeling. The timeline of the access to say, Why is this drop in the Axis? Well, there's not really any drop there is that it's upside down. The reason we can fall for this is because of the principle of convention that we've talked about a few times at this cost. So we just expect the Y axis the left vertical access to start at zero unclimbed steadily. So you can imagine how many people sort of glanced passing Lee this graph. They didn't study it or anything, cause, you know when it's posted online, Shared on Facebook. Put on Instagram, perhaps shared in the news. These kind of sites. No one's really sitting down and studying it. So in passing, they sort of reinforce a message that, oh, we seem to have less gun deaths in America. The results of this law not really in the case. Another problem with this graph is and not the key issue. Obviously, the key issue is that it's upside down, but one of the over issues with this kind of graph. Andi, I recommend any time you see a graph published over a large period of time just goes from the nineties to present day in the 10.2014 when it was published, I recommend you consider population. Consider what events may have happened across the timeline that affect these numbers. So I'm sure since 1991 to 2014 I'm sure Air Florida had some pretty significant population changes that my influence these numbers, I'm sure the police, Andi, government crime control and law enforcement. I'm sure there were some changes to law enforcement that possibly may affect the number of deaths in this graph. So what is essentially doing is showing you one number on a graph, the number of gun deaths on then giving you one single event and attributing everything to that singular event. So the reason the number of gun deaths changes year on year is entirely attributed to this point, which is not really true. So whenever you see a graph like this online, obviously check of its upside down when that goes without saying, but also anything that has multiple years especially, well, one to do with people or number of events that affect individual people. So if taking of a grain of salt, what think about whatever events may affect those numbers? So this is overall a terrible graph. I rated F minus for just blatant attempt to mislead, completely flipping Theo entire narrative of the graph literally upside down. And then even if the graph is the right way up, it's very misleading because you're pitching something that happened all these years ago and then attributing it to a single event on the graph. The reason the graph goes back in such a large timeline is probably the sort of demonstrate that the graph numbers change so much so they had to go all that way back in time to then present some sort of twisted narrative where the number of gun Jeffs fall, which is what they're trying to present him. Whereas actually, they rise Anyway, this is graft number one. I rate it f minus on my tips to you. Obviously don't post any graphs upside down. And secondly, whenever you see a graph that affects population, consider the population over this time period and consider over events that may impact the numbers you're looking at. Okay, so here we are with the graft number two in the Hall of Shame. So if you need to like the previous one, take a pause of the video. If you want to get familiar, the graph. But let's just take a look what we're looking at here. So the context of graph is it wasin the time off Brexit after you after the UK left on, there was a huge deficit in the U payments that UK left on. This is a graph showing you what the remaining nations of you had to pay. So to be a member of the U Here are the payments. They want a contribution off 1% of GDP of the nation and then underneath. Whoa. They want so much more money now, are there demanding a huge increase in the amount we pay? That seems to be at least a 45% increase. So wait, hold on a minute. It's actually labeled 1.8%. So what you're looking up is the difference off 0.8% not even a single percent off a bar. Yet one bar is almost twice the length of the evil one. So I think it goes about saying, I think you can see why this graph is misleading, Andi, because there's absolutely no scale. There's no scale a tall on these graphs. So essentially what they've done is presented this as a botch up. Except there is no bar chart to be seen here. These are just two red blocks. The each represent any size they want. So they just pasted one huge bar on one much, much smaller Bob, even though if this was a basket botch up on, it was properly scaled, these bars would almost be indistinguishable in size. So that is how this guff is misleading, so you can imagine the sort of viewer who might see this looking and at you with, if whatever their belief is looking at Brexit results and all that sort of stuff that goes into these decisions on day could even be totally swayed in favor of their belief or totally dissuade to what they actually think the situation. Whatever your view is on the whole thing, then this graphics setting misleading and it doesn't say a whole lot of anything. So this is an extremely common tactic, employed in newspapers as well as any of us that have media outlet or online social media or any anyone attempting to mislead, they tend. You'll come across this example all the time. It's when they show two or possibly more bars with absolutely zero scaling or with the truncated access. So there's probably a little bit of both going on here. What you're seeing is just one bar, absolutely massive on one bar, really small, so they attempt to mislead you with the size difference between two things that they want you to compare. Then who cares what the bus see? Let's just draw our own bars of any size we want on. I'll put it down and passed it off as a bar graph. So that's what they're doing here. And for that reason, I give this one and F minus for data misleading fact. I'm just giving them all F minus because no one should be attempting to mislead people with data like this. But hopefully I recommend to you that any time you spotted Bartsch up in the news, I generally go the rule immediately. Dismiss it because it's probably inaccurate, misleading or isn't, at the very least isn't telling the whole story of it, but I think this is a great example of just the bars represent nothing really there just to bars. They've put on this page on, labeled them and passed it off as a bar chart when actually they're just two blocks of red color. They've placed on the on the page and determined themselves how big each bar should be in an attempt deliberate attempt to mislead the audience. So here's the next graph from the Hall of Shame on this one seems to be showing the straw poll results from youth drawing the American 2020 election votes. Oh, it appears that Young is pulling far ahead of everyone else. He's got a huge part in level off the votes, massive pull ahead by Young and then everyone else is falling so far behind the amount of votes he has. Oh, wait, that isn't actually the case. So if you can see exactly how this one is misleading you, what they've done is attempted to show how far ahead a new young is on the results, whereas the reality of it isn't as striking. So you can see if you look at the percentages. Andrew Young takes 22.5. The next runner up is burning with 21 so that huge difference between them represents just 1.5%. And if you also look down further below, we sort of see underneath. Bernie, we have 17.9% which is do some quick. Remember, about 3% there, 3.1 on. You can see that the difference between them is actually far, far less than it is between Bernie and Young on you. Conserve. See that play out across all them. So again they have done something very popular him. This is one of the most common ways to be misled online. So if ever you see a botch up, I would immediately sort of dismiss it or be pay very, very close attention to the results here. So if you take a look at my version of the graph, so I've recreated it with actually the right scaling this time and you can sort of see the difference between young and bony hair is far less pronounced than it is on the one, whether trying to mislead you and show you how much popular more popular young is in this result. So this is extremely common way they do, so it's truncating the access on eliminating the scaling of the bars again. These are just images off bars they've placed onto this picture on, then just attributed values to them. There's no scaling. There's no axis. It doesn't align. But they tried to pass it off as a bar graph. So why, Why this is so dangerous is because even though I've showed you my graph, Andi, I have explained why Theo, the difference between Young and Bernie isn't as pronounced. That, actually is as it appears in this graph, is that even though I've explained that when you go away, you still sort of take away that Young was quite far ahead. So even though it's been explained, an audience will still massively misjudge the actual difference between the values, even though it's been explained that these values aren't accurate. So even though you can picture what the difference between 2122.5 is, you can't accurately visualize that because the visual display they've got here, even though inaccurate, even though you know it, you'll still go away and misjudge the difference. So they post these grass online, even though they know they're incorrect, even though it's obvious if you just take a second to look at the graph, how incorrect they are. But they still mislead, even though you know that you're still less likely to judge the graph accurately and truthfully. So that's sort of why these kind of graphs of really dangerous and misleading and just immoral in my opinion. But anyway, I give this one from the Iowa secretary of State and F minus for deliberately trying to mislead Onda doing so Paulie on pulling as many tricks as possible, even the formatting and the way this craft is presented as of draws you to Young's result so F minus for this one. Andi, I'll see you in the next misleading graph. So we've looked at a few examples off how they manipulate you have graphs in the scaling off bar charts. Let's take a look at the noble way in which they attempt Teoh intentionally mislead you with data. And that's to play around with the Y axis or any of our access in general. So let's take a look at the average annual global temperature from 18 80 to 2015. We can see while there's extremely low variants in the average temperature, in fact, it's remained consistently the same temperature all this time. But hold on a minute. This'll y axis goes from minus 10 degrees all the way to 100 and 10 degrees. Remember, this is in Fahrenheit, not Celsius as well. So this they've given such a huge, huge Weiss access scale to intentionally make the line look a lot flatter than it actually is. So, by a view, pluck, no matter how varying your data is, if you plot it on a y axis, which is long enough, then that data will look entirely flat. So the worst part is whoever made this governor exactly what they were doing. When they presented it, they plotted the average temperatures, saw that there was an increase, saw that there was some variance. And when that doesn't really line to what I'm trying to say, Hey, so let me just plot far, far more why value? Why values in the access that we actually need to do this data until the line looks suitably flat enough to misrepresent what the date is actually saying and present my biased view off it. So that's what we're looking at here. It's a very common one and a very big one in terms off populace popular, politicized science subjects such as global warming and things like that, where the topic is something that is very politicized and where the discussion relies on a lot of data. Then you can be sure that whoever is publishing it is probably trying to mislead you whatever side they're on. So I think my recommendation is any time you see a graph for any sort of issue like this, my recommendation is to just ignore it. It regardless off where you sit on the unfortunately politicized debate. I would recommend you take a look at the data yourself if you want to, or go to verifiable sources. Not just believe what is online what you posted, especially on topics like this. Because, like I said, it's a really common practice to manipulate the Y axis to flatten a line. Or you can add a huge amount of variance by zooming right in on the line. Andi on issue like this sit somewhere between those two values. But anyway, don't do this on your graphs, obviously, and also any time you see one online, especially about a sort of science E issue that has a lot of politics revolving around it, just avoid it. Take it with a grain of salt whenever you do so under turned to more reliable sources. For information like this don't just taken on face value. So with that in mind, I give this one another F minus for misleading Andi definitely intentionally doing so. When you have made as many graphs is, I have you can immediately just by the formatting and the way the lines are drawn, I can tell exactly what tool this was made in and this is quite clearly an excel graph. This someone has plotted the states into excel and made this graph. Andi, I know about excel is when you paste the values in the Y axis. Scaling by default narrows it down to the date ive pasted him. So in this case, it would be by default. The scale would be somewhere around 55 to 60 as the default value. So this person has gone in and manipulated that intentionally to make the line look a lot flatter than actually is. Now you don't have to have asthma. Much nerdy geek knowledge is me to be able to tell exactly what tool different visualizations were made in. Like I said, any time you see an access, why access where there? The point of it is how flat the line is, or the point of it is how variable the line is. Then take your of a grain of cell that that probably in trying to mislead the data to represent it the way they want. So I'd like you to take a look at this next graph which misrepresents data on goes in my hall of shame. So here we can see the UK national debt on what we're looking at. The graph is from 1995 to 2016 and whoa! It seems like the national debt is increasing with unheard off results like we're reaching brand new highs that we've never seen before. And this is possibly due to the political party that is in charge at this point in time or the climate we exist in, We've got huge amounts of GDP. It's spiraling out of control. It's almost going off the size of the graph. It's so large, in fact. Well, this is a case off omitting data, not cherry picking data. Too much the intend, the result you want to show rather than the actual results. So now I'm gonna show you three full graph if we zoom out a little bit. If we had a few more dates to this graph, what does it? What does it look like now? So here you go. This is the same graph the same measure. Only this time the timeline has been extended to 1910 so we can see a lot more years. We have national debt till a cap on did actually tells a surprisingly different story. I mean, I'm shocked that the up for this graph would attempt to skew some of the data. So what they've done is they've picked a portion of the timeline that marries with the intended story they want to say, rather than the truth for one. So they've gone from 1995? Not because that was where the best represented data, but because that was the all time low. So they picked 1995 to show you the room the lowest point on the graph, so you can start from a low point on, Then you can see the increase as asses they want to show you, so they want to try and show you how much is increased. But if you actually look at the full data set, the this increase was seeing on paragraph is actually do owe an overall downward trend. It's a small bounce back, a small rise in an overall massively downward trend of the graph. So you can see how they've picked these data points. I intentionally to try and throw you off. So this is a case of where the data itself isn't lying. It is old truthful data on the story they're trying to present isn't one that is fully represented in that day to set because they've cherry picked it. They've thrown out the results that conflict two of what they actually wanted to say on only handpicked the results that conform to the message they're trying to say. So it kind of goes back to one off the air points, I said earlier in one of the Hall of Shame graphs is whatever you're looking at, something over a large period of time, consider the factors that affect that graph because we're something as complicated as national debt. You can't really tell a story of a few bars on a bar graph. It doesn't really say everything that all of the factors that contribute to that so usually when they cherry pick things about, they sort of toss graphs around to align to whatever the messages that they want to tell you, as opposed to one that so that gives you a full on clear picture. So again, if you see graphs over large paid of time, consider things like population. Consider things like events that may have affected the results of this graph Because something this complicated issue can't present in just a bar. It's not really fully accuracy to the full story. So I recommend to you any time you see large graphs over long page of time, consider what the data might have been before the timeline, where they've started to show you the data on consider what other factors my go into the results. You're actually looking up because usually grass like these, they don't give you the full picture. And so the next way that people attempt to mislead you with data visualization is the sort of infographic where I would say they've tried to add a little bit of innovation to the way they present data which you'll commonly come across in infographics where what, you're looking up just it doesn't make any sort of sense wise. It presented this way instead of any traditional way. Well, they could be trying to be a bit fancy, sometimes attempting to be fancy and deuce doing so I inadvertently although it's certainly a practice when they're attempting to mislead, is to introduce some new, innovative way off presenting that data that ultimately is misleading and they've done so intentionally. so the sort of innovative element overtakes your mind and takes full front rather than the actual misrepresentation of what they're showing you. So here we have and a timeline. I guess you could say this is what this is supposed to be off the different owners or presidents or CEO or head roll at the Papa John's pizza chain. And you've got the names off the different leaders on the timelines they served in represented innovatively as a pizza. Whereas I think it's well documented. My hate of pie chance on even when you use them as a actual pie or a pizza only adds to how much I hate them. But anyway, this graph, if you look closely, none of the slices represent the actual timelines. They're entirely misrepresentative. If you start on the left, you see Johnny's got a huge pie slice there, 50% or whatever. Almost. It's hard against hard to tell with a pie chart. I don't really know how big that slices with any accuracy, because that's the problem with pie Chance. I just can't tell. And then you next to that you've got Nigel who, if you look, represents three years but If you look at the first pie slice that represents 15 years, so that's a big jump, and then we have the next 1 2008 to 2017 0 based on this pie chart, that must also be three years because it's the same size as the previous one, which only represented three years. And then we've got one year slice followed by 2018 to present, which, um, I'm not entirely sure when this was posted, but even if it was today 2020 that could only possibly be two years, and it's a very thin slice on. We also have a completely unlabeled pie slice. So who knows what this graph is trying to say? I don't know why they've picked it. I suppose it was attempting to be fun, but also the same time. Just I'm trying to show how much of the timeline Papa John himself takes up. So this is, ah, bit of a fun one to show you, but there are a lot of underlying element. So you try and you see something like this, and you can imagine a viewer getting distracted with how phone and innovative something like This is where, as the reality is, they attempted to mislead you. Or sometimes they do possibly not in this case, but often when you see infographics and fun visually toe look at. But usually that doing something a bit funny with the numbers on there because they're encoding values into pictorials or, in infographics themselves, a little icons and things like that. Where is no graft? Old does that for you. So you have to rely on the actual artist or whoever the author is publishing this being entirely truthful and measuring ALS the different angles of the pie chart, pizza that they've made, or any time there representing infographics. You're relying on them to do the scaling. Which again we've talked about is to problems is when you're trying to scale something by heights. But it's a two dimensional image than you actually scaling it on two times, which means it triples in volume as it increases in height, which is very misleading. Unusually infographics distract you with something cool some innovative way, and you end up with something like this. The big cheese, which no one's labeled all the segments correctly. They don't represent anything on ultimately and they the pie slices. They've just determined how big they are. They just put pizza slices hair, decided how big they are and then labelled them like it's kind of some kind of data representation when it's not really, and you'll see a lot of this in, especially when it's a company publishing their results or publishing their activities in a deliberate attempt to mislead you. So you kind of see on a very popular example would be different cigarette companies showing you benefits or health detrimental effects of their products. And they've done a cool infographic to show you the truth behind the numbers, whereas they've done something funky to it, such as hair. In this Papa John example where the slices don't actually represent a thin, they're not actually scaled. You can't understand what they are. And in fact, not all of them have even been labeled. So usually don't turn infographics for sources of information or data. Try and look around the infographic. Sometimes they post the source of the data on. Then click on that Lincoln going Look for yourself is the infographic truly representative because an infographic is something that someone sat down and drew themselves on the representing icons is encoded values, whereas were totally relying on them to just draw them the right size. On with this pie chart, I doubt anyone actually start there and got a protractor out and measured the angles to make sure that it's truly reflective of the numbers. They just put the part pizza there, randomly, sliced it up and then labelled it. So if you're turning to infographics for sources of data, although fun to look up are probably misleading. So with the final graph in my hall of shame, I present to you this monstrosity. So nothing particularly misleading here. I'm not going to be talking about how it's trying to misrepresent the data. I just wanted to share this with you and have it documented so we can all see the appalling nous off this graph. I think, even though it's lost its labels as I'm not sure where the sauce that this one came from, but the original sauces somewhere the labels have been lost. But you don't that they're not going to write anything to this graph. It's still totally illegible. I think what we're actually looking at is a scatter plot. So we have the two dimensions of a scatter plot where instead of bubbles or instead off dots, they've represented them with pie charts. Then they've scaled those pie charts across two dimensions, so they've added another layer of data density. Then they've added a vertical access to those pie charts on they are where they call it stacked pie charts of this pie chart a top pie chart, the top pager on. Then they've divided those pie charts into the different segments. So you're possibly looking at eight different dimensions of data that they've attempted to represent on this graph. I mean, good on them for EMP adding some data density but doesn't tell anywhere near a coherent story. And I'm not even sure where you how you could even understand this and why anyone would make this and present it. So just a little fun one at the end there. There's nowhere misleading to discuss in this graph for how intentionally misrepresents data because it doesn't represent any data toll. It's totally unreadable. So I'm awarding this graph. If all of the over graphs were f minus is this one is so much worse than I'm gonna have to give it a zed minus for effort on the zed minus for overall representation of data. So possibly the worst graph I've ever seen. I came across this, um ah. While ago on Guy just saved it. I just saved it to a folder, thinking one day I might reflects back on this graph and have used for and this putting together this cost came up. So I thought this would be a great thing to show everyone. So if there's one thing you take away from, this course is, don't make anything like this. So there it is. Possibly the worst graph in existence is right hair and you're looking on it. Great. So that's been the hall of shame on all the different ways in which media can misrepresent data or people can misrepresent data to try and intentionally mislead the audience or sway them one way or the over in an untruthful manner. And I think what kind of should frustrate UAL is how intentional it all is. None of these graphs were by accident. They were all deliberately created. This way to misrepresent what they're actually saying is which is just in my opinion, there's an a moral component to that, but it should frustrate. Anyway. I hope I have a quick new with the tools necessary to sort of spot misrepresentative graphs . You mace come across from time to time. Just avoid being misled on. Always turned to legitimate sources. A true fear data. Andi, I assume as you're taking this car, shows someone who enjoys analytics, someone who enjoys data. So I would recommend you go out there, look for the source of the information and put together your own graph. See what it actually says yourself. Okay, great. So this has been my whole a shame. Now let's jump into the next section where we're gonna learn how you actually create some really impactful meaningful graphs that resonate with an audience. Thank you. I'll see that. 17. Remove Distracting Elements: hi there. A clear graph is essential to communicate a clear message. If your audience is struggling to understand the message of your data visualization because it contains too many distracting graph elements, then you won't be communicating with impact. Most data visualization tools by default, such as Excel Power, BI I or Google Sheets, their default visualizations have too many distracting graph elements, which need to be cleared away in this section of the costs were gonna learn how to spruce up and spring clean your data visualizations so that they tell a clear and poignant message . Firstly, we're going to take a look at exactly what visualization clutter is on. Why it's important to clear. Secondly, will then look at the steps you need to take in order to clear it up so that you can have a clear direct message without removing to much of the graph so that the message ends up diluted and lost. So what is visualization, clutter? Visualisation clutter is anything that distracts from the key message off the visualization . All the graph elements that you find on a graph can be considered clutter if they don't directly contribute to the message each and every visualization has a message that is trying to convey to the audience. It could be that sales are doing well this month, or perhaps that you've experienced a peak in employee attrition. Either way, there's a message that the audience needs to understand, and the visualization communicates that message. If your communication is cluttered with Graf elements, this makes the audience have to stop, pause, read through the graph and then try and interpret what that message could have. Bean. This is an ideal at best. You add confusion. The audience takes time to understand the message, and they begin getting distracted. And at worst they could interpret the message differently, nor incorrectly, because you left it open to interpretation. Every visualization is composed of the graph elements the lines, labels, titles, access names, etcetera, all elements of the visualization. Each time one is used, it could potentially distract from the message that you want to communicate to your audience so you must review each and every element that composes the graph individually, And if the message could be understood without that element, then you probably didn't need it. It just adds to the clutter. The point of this step in the design process to an impactful communication, it's to create a clean and clear graph. We're not gonna be making any formatting choices at this point. We're just gonna be setting out all of the components of the graph that we actually need, ready for them to be formatted. So they align with the visual design principles that we covered earlier. With that in mind, let's take a look at an example. Here we have the sales revenue generated from three cinemas over three months. This visualization contains many different elements, and not all of them add to the clarity of the visualization. So let's take a look at each of these elements and see if they add or detract from the message. It's good to get into the practice of reviewing each of them one by one. And as you get more experience in creating graphs, you'll begin to get a sense of which elements add on which detract from your message. Ah, habit that I have formed is to delete all of the elements and start with just the bars themselves, then reintroduce elements until the graph can be understood. Firstly, the Axis labels these the labels written on the access that let the audience know what is being measured by the visualization. In this example, we have to the numbers representing the some of the sales revenue Onda along the bottom. We have the mumps without the mumps being written on the axis, you wouldn't know that each of these three bars represents a month, so this adds to the clarity. Therefore, we're going to keep it. However, the access that shows a sales revenue found along the side of the visualization doesn't add anything to the clarity. We already have data labels with the values, and they're much clearer. Repeating those numbers only distracts from the message, so we're going to remove them. Next. We have the key, which is used to inform the viewer of particular elements and their meaning. Here it is telling us that Feb, March and April represent mumps. In this case, your audience is going to understand that these are actually months, so we don't need to repeat that to them. In this case, it just detracts from the message and is considered clutter. So we're going to remove it. Next we have the data labels. These the values that you confined, pasted on top of the data elements themselves that let the audience know what each part of the bar represents. In this example, if we remove them, you wouldn't know how much each of the months generated in revenue. So they do add to the clarity, and we're going to keep them now. You could have also communicated the value of each bar to the audience by instead of using data labels, you could have kept the access which communicates to them the size of each bond. What that represents. Personally, I find that data labels are a lot more clean and clear and communicative to the audience because the place right on top of each boss segment, you're told as soon as you look at the bar segment what the value is without the need tohave to constantly look left and right and scan each part of the bar and then look over to the left side of the access to see what that value actually represented. So in most cases, labeling the day to themselves with the data labels is best practice, and to me it looks a lot cleaner. Next up we have the grid lines. These the lines of the back of the visualization. You tend to find one. Each of the Axis label points by default. Most visualization tools will contain grid lines. I find that they're simply not needed and only add to distracting the audience from the actual message. So what I'm gonna do is I'm going to remove them. However, you probably noticed that a lot of the graphs in this cause that I created contained them. I find with grid lines, subtlety is best. You can even remove them entirely. Or, if you choose to keep them, make sure you pick a very neutral shaded color for the background. Then you can have subtle white grid lines like you'll find in most of the previous visualizations I find. Doing that creates a clear separation between the foreground and background. It makes the date elements stunned out a bit more, but that's just a personal design choice. Keep them or remove them. Either way, keep them very subtle. So now let's take a look at the legend, the legend. That's further information to the audience. To help them understand the visualization. Here we can see the legend is informing us that each color section of the bar represents a different cinema. Later, in this course, we're going to take a far deeper Loggins how to create effective legends and how we can use color. So right now, the legend is adding information to this graph. So I'm gonna keep it there we have it. We've completed a very small and simple by entirely crucial step to creating an impactful data communication. So now we cleaned up our visualization and removed any of the clutter. This gives us a great jumping off point to make a visualization that's gonna be impactful and resonate with an audience. Ah, goal With this step was to distill the graph to just the essential components so that we can work on these elements and apply some visual design principles that we covered earlier in the formatting. As I've created thousands of graphs, I have a good inclination of which graph will best represent the data that I have. If you're exploring the different graphs and new to communicate any data, you may start with picking the best graph or start with a graph and just start clearing away the clutter. It's also pretty common to go back and forth between these two steps. As you're working on your graph, you may think of a better way to represent the data that you have. And after you've changed it from, say, a bar chart to a line graph, you may then have to reintroduce some graph elements to tell your story. This is totally normal, and sometimes I still do it. So don't always expect to pick the best graph on the first attempt. You definitely allowed to swap around with Graf elements to tell your curated story. Decluttering a graph is a pretty easy step. You just have to follow a simple rule. Review every part of the visualization, and if it doesn't add to the clarity, then remove it. Start from a blank point and add. So now that we've de cluttered our graph, join me in the next section of the cause. We're going to review in detail the use of color and call outs to really highlight this story within the data 18. Bringing out the Story with Colour: Hi there. Cologne is often overlooked, part of creating data visualizations. Sometimes the author ops for color palette choices that they like, rather than ones that add information and data identity to your visualizations. So in this section of the costs were going to be taking a detailed look at color and lend the do's and don't when applying it to data visualization. We're going to be going beyond simply using red and green toe indicate things are positive or negative. So join me in this section of the costs where we're going to take a look at all the ways in which you can enhance your data visualizations with the use of color to leave a memorable lasting impression on your audience. In this section of the costs will be reviewing all the different ways in which color can be used with data, visualization and design. To reinforce the story that you want to tell, we'll be looking at colors that support convention, such as reds and greens, as well as exploring how color config used to highlight aspects of the story to adhere to visual design principles that we learned earlier, such as order and relationships. So let's take a look at all the best practices of color as well. Some things that you should be avoiding. The first principle of color we're gonna be taking a look are is red and green, which indicates negative or positive things about your data visualization. This is probably the most common obvious way in which you can use color to communicate. Just think about how often you encounter these colors to indicate negative and positive in real life things such as traffic lights, bathroom stole lakhs, some electron ICS as well have charging in the haters. So when you plug in your phone, no one told you what it meant when the indicator light went from red to green. But you instinctively know that probably means that your phone is now reached full charge. You just knew this known how to tell this to you know how to explain what this indicator means. You knew it just from the use of color on the light. So this is an example off the visual design principle we looked at earlier called convention happening in real life, your audience will just instinctively, if not always consciously associate read as bad and green is good when you're creating data visualizations. You need to be aware of this when you're designing them. Often times. Some people might opt to use one of these colors as it looks nice. Or perhaps it's part of a company color palette. This could confuse the audience. Is the message you want to communicate? Could conflict with the colors that you've used. By using these colors incorrectly, you might have to overcome the audiences natural propensity of convention, which is that naturally gonna associate red and green to mean good and bad. And you have to overcome that when you're trying to communicate. Something is good, but you've called it in red. So don't fight convention like we mentioned earlier. Just go along with convention and red and green is a really common and strong convention through the use of color, and it comes up often when you're trying to design and communicate with data visualization . So even if your company color palette is red and green or uses those colors, try and avoid using them in your data visualizations, unless you are trying to communicate that something is good or something is but to demonstrate the use of these colors in action. I want you to take a look at this bar chart. Example. What it shows you is the sales revenue over a few different months, the presenter wants to communicate the April and may have a significant drop in revenue. This isn't entirely clear. Without being told directly, we can make this graph a little more easy to understand and communicate information quicker to the audience by changing some of the colors to red and green. So by changing these bars to green, red and even orange now, the presenter is communicating to the audience that each of these values is good or bad in some way in context, after business in the communication they're presenting now, beyond just communicating absolute positive or absolute negative. With the use of red and green, we can actually use it to add another dimension to the visualization and communicate additional information. In this example, the lowest bars have been colored red to highlight Par Salesman's, however, suppose in the context of this business that they sell a seasonal product in April and May actually quite expected to be very low salesman's. So it isn't a surprise that they are the shortest bars. We can actually add to this visualization by using these colors in a different way. Now consider the fact that the business set sales targets each month. Using these colors, I can add this dimension to the bar chart. For example, here I've used red and green to indicate if the sales target has been reached for that particular month. See how the understanding of this graph now completely changes April in May, a low sales moments in terms of total revenue. But since this seasonal products, it's expected that the sales revenue is going to be low. However, they still exceeded their sales target. Now no longer is the low bar, indicating a negative sales a negative thing. It's quite a positive thing now because it's been colored green, where some of the taller bars have been colored red look at February and June. In the previous graph, the green color lead you to believe the high sales these months achieved was a positive thing for the business, whereas in this context it isn't actually true, even though they are high salesman's naive of them achieve their sales target, which is negative thing for the business to see how using red and green in different ways can communicate different things. With this bar chop, you can add more data identity. That's the tone we use when we add additional things to the graph and indicate different bits of data to the audience for just the use of color. I've now added an additional dimension to this bar chart on. That's because people follow the convention off red and green to indicate things to them. So I can very quickly communicate that to an audience without needing a second graf for adding labels to the boche out that indicate which months exceeded or fell short of different sales targets. If you think back to the original blue graph, you aren't sure what the key point of the graph is without using color. The offer has left it up to you as the audience to review and interpret the graph in your own way. There's no way you'd be able to tell which of the mumps reach their sales targets and which ones didn't or which of the months or a positive or even negative thing in the context of this business and this communication. Now look, bucket this graph. It's immediately understood by the audience that Feb in June or focus MMS Is there something negative about them In this example? It employs design in the visual perception concept of convention to quickly and effectively communicate a custom message to the audience. But I want you to keep in mind when using red and green color to indicate a positive or negative thing in your data visualizations. You don't have to use it just because it's there. You have to consider the audience on whether or not this information is important to them in the objective of the communication. So don't just use it because you have these tools available. You don't have to label everything good green and everything bad negative only if that's the story you're trying to communicate. This could be a powerful tool to add additional data density or very quickly and effectively communicate things in the context of the business that are negative or positive . Color could be used sparingly. Data visualizations. It really is true that less is more when it comes to color in data communications. So now let's look at another way in which we can use color to communicate effectively with our data visualization. We're gonna use color to highlight importance and create an order in the visualization, so that makes it easier for the audience to understand. Recall what we learn from the visual perception concept of hierarchy. If you want to brush up on that before we start, you can head back to that section. I'll be waiting just here. But as a quick recap, this concept refers to the way in order in which the viewer reads and understands the visualization. Remember, when you view a data visualisation, the audience isn't looking at the entire thing it wants and immediately absorbing all that information and coming to an understanding of the graph for the story while they're actually doing is looking at each and every individual components separately, one by one, sometimes randomly. Across the visualization is the eyes dart around until they come to a full understanding of what they're looking at. Your aim, as a data visualization expert, is toe help. The audience come to a quicker understanding of the graph by making it so that the important parts of the graph for highlighted and grab attention immediately. This could be done through the use of color up to highlight important parts that reinforce the story as well as creating a foreground and background. So it compartmentalizes the graph that makes it easier for the audience to come to a quick understanding of what they're looking out. We interpret things, is a foreground and background. We've objects in the foreground taking importance, and they're generally larger, bolder, brighter objects that we place in the foreground. So knowing that we can take advantage of this and use color to know only aid our understanding and quickly interpreting a data visualisation. But we can also use it to communicate this story to the audience as an example. Here we have a graph showing a company's attrition over a period of six months. Look at this graph and see if you can identify the ways the visual design concept hierarchy is being identified. I'll pause it right here if you need to, and just write a few notes in your workbook. What I want you to notice from this graph is the two identifiable layers. There's a very clear foreground in a very clear background. The white space of the graph is the background and the faint grey lines in the acts of the access and the grid lines and the muted grays and small funds used in the access level push these elements into the background as well. But the bold and the large funds of the title and bottom axis labels on the use of bright colors in the line place it so it clearly sits in the foreground. And then this is what grabs the audience's attention. These a deliberate design choices that have been applied to the formatting of the graph, the's design choices aligned to the visual design principle of hierarchy. In order the aim of thes design choices. It's to create a foreground and background and aid the viewer in reading things in in order that helps them understand the graph. The parts of the graph that are important to understanding have been placed in the foreground. The title The Axis labels on the data elements themselves are now what jump out to the audience first, and these are the important components that take attention of focus, so the viewer will come to an understanding quicker. The less important parts have been placed in the background is they don't aid in the understanding of the graph, the's design choices, how we can make the graph more eligible and easy to understand, which is great. But the story of this graph doesn't stunned out yet. The graph itself is an effective choice to visualize the data, and the clutter has been cleaned away. Andi, we've used color and designed to adhere to the principles of hierarchy in order. But when an audience member views this graph, even though it's clear it doesn't communicates a message, there's no story in this data visualization. The View Est still needs to interpret the data on come to their own conclusion about what it means. So now this is when you need to bring out the story in your data visualization. We can also use the same principles of design toe alter the hierarchy to communicate information to the viewer. Suppose the author of this particular graph wants to communicate to the audience about two months, in particular, as they want to drive a conversation about why these two months were different to the rest . In this example, the company's downsizing from May and wants to talk about the increase in attrition without employing the use of color and designed to highlights and create a hierarchy of layers. This story doesn't stand out to the audience, so let's see how we can now use color to bring out the story in this data visualization to now take a look at this scene graph that communicates the same information, but I've made a few additional design choices here to really bring out the story. By using color and size, I have shifted the hierarchy of the graph tow line to the main message. I've drawn the audiences attention to the two particular mumps relevant to the story may in June. This was achieved by using a brighter color on the line access labels on Axis labels for those months. Also, the only if I highlighted the two months of a brighter color. I've pushed the Outer Moms into the background by shading them a lighter, less noticeable color. The's design choices don't just create a graph that it's clear and easy to understand. It's created a graph that tells a story instead of just placing elements that aid in the understanding of the graph, such as a title and the data visualization elements themselves in the foreground so that they take attention. Now we've placed a story in the foreground and everything else on the graph in the background. With this just a glance. An audience is able to tell that these two small bumps stand out in some way is they're important to the context of the story that's being told here. The reason why they stunned out isn't yet clear. So in an upcoming section we're gonna be uncovering How weaken bring the story out fully. For now, we're just gonna be focusing on design and use of color in this section. So we've learned that color and design choices that you make on your data visualizations can be used to create a hierarchy or follow an easy order. This allows the audience to come to an easy understanding of your data visualization. We can also use color to highlight the important parts of the data visualization that directly contributes of the narrative. You can also use conventional colors to indicate if things are positive or negative in the context of the audience or the business. No, the way in which color can be used in design is to create comparisons between values. If you present graph to an audience, and two components are colored in similar colors. Then the audience is going to take these as related, and then make a comparison between those values. You need to be aware of this. Firstly, in case the point of your visualization is to show how to values, compare or so you don't accidentally create a relationship between values that shouldn't have a relationship. Let's take a look at this graph that shows the number of hires for hay HR department to just take a moment to get familiar with this graph. You compose the video if you need to. Oh, just take a look at the graph in the Hunt Out book before I reveal to you what the bottom axis represents. Let's take a look at how the use of color can communicate different things to the audience . So what we have here is to Siri's of bars. We have a series of blue bars on. We have a Siris of orange bars comparing the blue bars across each of the Siri's. You can see there's a variance so you can see the blue bar starts up. Then it dips down in the second blue bar, and then it goes back up again, reaching a new heights in the third blue bile. Where is in the orange bars? They steadily increase the first orange bars quite low. Then there's a slight increase, and then again, there's another slight increase. So what do you think the bottom axis could represent? The bottom axis could represent somethin like tying. Perhaps it represents different months, and the two bars in each of the months could be the value of this year compared to the same value of last year. For that particular month, you'd be correct to assume something like that is the coloring the bars makes you compare them. The way these bars have been spaced and colored invites you to think that the graph represents to Siri's of values over three periods. Because of these design chases, you don't assume that it's actually six individual data points or three Siris of data. Over two periods, they've been grouped together in pairs and colored. As such, these design choices it hair to concept of relationships in visual design. The proximity and color choices creates certain relationships between the data points. If this graph wasn't actually about to Siri's of values over three individual points. You wouldn't be able to get an audience to easily understand that. So suppose the bottom axis represent three different recruitment team's on its showing your comparison between the number of hires against the target. The design choices of the graph lead you to understand that there is to Siri's of values, which conflicts with the reality that there's actually three distinct categories. That's one team second team and 1/3 team. And then we're comparing two values within each team. So now on this graph, what I've done is I've made some design traces that adhere to that story by coloring each of the three different teams a distinct color and then the both bars of in each of those teams a similar color. I've now invited the comparison to be done accurately. The audience can now see there's three different categories and two values within each one . To be compared to this graph is exactly the same as the one we looked at before. All I've done is adjusted the color notice how the message of the graph entirely shifts. No longer do you notice the second bar steadily climbing, but instead you can now see how teams one and three hide more than that target going. Each Siris of bars separately, like shown in the previous graph, conflicts with the message. This leads to confusion is your audience is going to compare each value in this stare. Ease instinctively because they've been colored in similar colors, even if you tell them not to. You want your coloring to reinforce your message, not compete with it. Bye, Miss Coloring You maybe suggesting comparisons to your audience on unrelated figures, which makes the message or story of your communication on clear and confusing. Whilst you're talking about one set of figures, you audience is making a completely different comparison. Now that we've used colors to invite the audience to make the correct comparisons, we can hone in on the story by combining design methods. In this example, Teen to actually didn't reach their recruitment target, and that's what the author wants to present to HR management. You can actually combine different coloring methods as teen to is the focus. I'm gonna make the colors a bit brighter on light in the colors of Team one and three to reinforce the focus on team, too. Now you can see that the bars have been colored impairs to reinforce the relationships between them and the boss of team to have been brightened to focus the audience's attention on the story of this visualization. In this section, we're going to look at how we can use color scales and apply them to data visualizations to reinforce or communicate a message. Morgan alone when best to use them and went to avoid them. Firstly, let me explain when a color scale is simply put, it's when the color of a visualization value changes from one color to another or a long ingredient scale to depict the value. Okay, that's a bit of a technical description, but in simple terms, it's liking this example. Where is the value increases? The color changes along a scale from red to green. It doesn't have to be from one color to another. It could also be from a lighter shade of a color to a darker shade of a color. Let's take a look at an example. Here we have a table from a call center. The first column shows the number of received calls per month and the second column shows the number of missed calls per month. I can use a color scale in this table. I can apply a scale from a dark blue to a light blue, depending on how many recalls were received on Mr of in each month. The light of the color, the lower the number. It's a great visual way to quickly interpret a list of data for ranking. Similarly, you could rank everything on a scale not just largest to smallest value, but good to bad. By using red to green scale, you can also use color to scale as an extra dimension to the visualization. For example, here is the scale of the average customer feedback score at this call center. The darker the shade, the higher the feedback. You can see how June and January are opposite ends of the customer feedback scar, but have the same amount of received calls. So by using the color scale, I've added another dimension of data to this visualization. The scale doesn't just apply to the values being perceived by the audience. They add an entirely different value. Color scales are a great way to quickly communicate a dimension of data to an audience. But, like with the red and green, you shouldn't just use him just because you have them, there is an appropriate time to use them. The best place to use a scale is when there is a lot of data to consume, and you want the audience to get a quick understanding, such as a table or scatter plot, where the color adds another dimension or allow us a quick understanding of a large data set. You also want to avoid using a scale on a visualization over time, such as a line graph. It just doesn't really work to wreak up. There's three main ways to use color one ad hierarchy to your visualization to aid in understanding by creating a clear graph with white space, a background in the foreground and focusing the audience attention where it needs to be to to communicate convention to the user, such as positive or negative for the use of green and red, and three to indicate relationships between the data values or to ensure the audience doesn't see a relationship that doesn't exist. So we've covered the main ways in which color in data visualization could be used to communicate data and information or a story to an audience, but that's not it. When it comes to color, there's a few best practices that we're going to cover, as well as some of the things you should be avoiding when using color. Firstly, the first thing you need to avoid is the multicolored rainbow visualization. Like in this example, this visualization just has too much going on. Every bar has been colored a separate bright color that's all competing for the audience's attention. Some of them follow the convention of red and green. Some of them don't. There's just too much color going on in a graph like this. The aim with color and data visualization should always be a deliberate choice to communicate or reinforce the story. You shouldn't just apply color for the sake of a playing it. This author has tried to follow some of the principles. For example, each bar isn't related, so they've used a different color for each bar to ensure that they're separated so that a false relationship isn't made by the audience member. They've also used bright colors for these data values to create the foreground and grab attention from the audience, and they have appropriately used muted colors to create a background, which is all pretty good practice. But because so many different and bright colors have been used, it's actually having the opposite effect than desired. They intended to make the botch up more understandable. But because each bar competes for attention, trying to grab the focus of the viewer to each bar independently makes the comparison between them quite difficult. On the comparison between bars is, after all, the main point of a bar chart. So they ended up creating something that had the opposite effect and desired. Instead of making it easy to understand, it's actually now a lot harder to understand. The best practice for the color palette for your data visualization is going to sound a little barring when it comes to all of the faded background elements of my data visualization. I go for a very neutral color, such as a very muted gray, then my go to color for data visualization elements themselves that I need to make bright and focus is a blue or teal color. And then, if I need to bring to do some different colors to sort of accent things. Then I try and pick a color palette that matches on doesn't have any convention to it, such as like a yellow or maybe a bit of an orange. So most of the time your graph should be grays and blues, which sounds a bit boring. But remember, you're trying to create an effective story out of them. It's not the graph colors themselves that do the story. It's how you prepare them and design them. So with grays and blues, the focus of the audience is gonna be on the actual story behind the data, and they're not going to sit there and go. And while this is a very pretty looking graph, because so much coal has been introduced and he ends up having the opposite effect. So my recommendation to you is to just stick with gray and blue for most occasions, introduce and green and red where required. And then you can also add a couple of different bright colors here and there, sparingly, such as a yellow perhaps. So. Now let's turn back to the same graph we looked at earlier. I've removed the bright multi colors on. I've colored it appropriately Now I've picked a nice blue or teal ish color for the data elements and everything I want in the background. I've kept a very muted, gray faded color. By changing the colors to be consistent, it makes it much easier for the audience to compare the values. But wait a minute. Didn't I say earlier that you should separate the values with different colors so that the audience doesn't accidentally create a relationship between them? Well, it's true, but what you should notice The's values are in fact related. They may be different stars, so you could believe that they aren't related. But what you want is for your audience to compare them. When visualising data the relationship is between the data points, you want the audience to compare, not necessarily the business context of their relationship. In the context of this business, each store is considered independent, but you want your audience to be comparing the sales between the stars, which means they are related in the context of the data visualization, so you should invite that comparison by keeping them a similar, consistent color. The next best practice is something we locked out a little earlier, and that is to avoid being tempted using your corporate branding colors in the graph elements. Many businesses use red and green in their branding your image, and then it could be tempting to use them in your data visualization. But as we mentioned earlier, this should be avoided because red and green carry contacts that you might not want to communicate your audience Beyond reds and greens. Many corporate logos and branding use colors that clash, which creates a very busy, hectic, tough to understand visualization. You should opt from matching colors, but don't worry. You don't need to be a visual designer with a degree in color theory. There's lots of free tools out there to create great color palettes. So I'm just gonna jump right now into the computer, and I'll reveal my secrets of picking great color palettes from my data visualizations. Hello, Here I am at their computer, so I'm just going to show you my secret weapon to picking color palettes. I'm no air visual designer. I don't know a lot about how different colors it created and what makes a matching color on what doesn't. But when I do know is color palettes are really effective, and they just add that extra level off professionalism to your data visualizations. So you should definitely avoid clashing colors and go with something from a color palette. So this is what I'm going to show you of how I pick a color palette when I'm trying to do some data visualization. So there's many tools like this out there, but this is the one I like. It's called cool Laws or Coolers. I'm not sure exactly how you pronounce that. I'll just show a link to you on the screen now on. It's also going to be in the hunt out book. So with this color palette generator, it does exactly what it says it creates. A color palette for me does so instantly, and it's totally free to use. So I'm just going to start the generator. I'm gonna click this blue button here, and then it's going to generate a random color palette, far me. So this tool uses some kind of really intelligent ai. Too much colors that go together automatically on what I'm looking for is I'm kind of looking for couple muted colors, a gray hair, maybe a darker gray sort of color that matches. I'm looking for my blue, which I'm gonna be using in my data visualizations. And then I'm looking for a couple accent colors. Perhaps so it's randomly generated me a pretty decent color palette already. This would make a great blue for some my data elements, and I quite like it. So what I'm gonna do is I'm gonna hover my mouse over it, and I'm going to click this little lot. I come. This means it's gonna lock this color in. So I like this color. I've locked it in now and then All these colors much with this blue. So I've already got a kind of slightly off white, which would make a great background color for the so, you know, to create the background of my graph. So I'm also gonna lock this Run him. I also kind of like this dark gray. So if I'm adding any text elements, or perhaps I have some access lines or something like that on my communication, this would make a great color for the funds or different elements where I need to do a bit of writing or use a dark color. I'm also gonna lock that one name. So I've got my three colors locked. But let's say I need a couple accident colors because I want to use a couple different bright colors to sort of create a different relationship between my data. I need more than just one color visualization. So what I've got with my colors locked, I'm just gonna hit space bar on. What this does is automatically generate some colors that match. So in this case, my pink here, I'm not quite keen on. And I don't really like this mint green, so I'm just gonna keep it in space bar until I get some nice colors that match. So I've kind of got a nice yellow here. So if I wanted to create a bar chart with Blue to represent one Siris of values and I need another color for second, Siri's a nice yellow like this would be a pretty good I'm not quite sure of how bright it is. So I'm allowed in this tool to make a few quick air sort of adjustments so I can click this one. I can adjust my shades. So perhaps I want a little bit of a dock sort of yellow color. So I'm gonna select this one, and then I can also click on this just and I can adjust the brightness. So I'm just gonna tweak the brightness, maybe a little more such a regime. And then I kind of got a really nice color. That much is my color palette. I've got a really nice color palette going here. It's even given me of red that I could use if I wanted to communicate in context and convention some negative values, that much along with my palate. So I've got a pretty decent pilot hair that I can now apply to my data visualizations. So let's say now jump back to the presentation. I'll show you kind of how these colors have come together to create a graph. And when you're creating your different elements, depending on the different tool you use, you may use hex values. So here I've got the hex code off this particular color. So not all visualization tools rely on hex. Sometimes they used the RGB. So what you need to do is you just need to copy this hex code and then Google some kind of tour that will change checks into RGB. Then you can start applying this color polity of graphs, so I'll leave you. Now I'll jump back over to the presentation on. I'll show you a couple graphic samples where this color palette has now been employed, and you can sort of see how it all comes to. Gavron automatically matches. Great. I'll see you that okay, here we are back at the video that you can see. I've put together this little data visualization on story, using the colors we just generated in the color palette, so we don't forget to detailed in the story. But it's just a bakery. It's showing you a couple of the products that bakery sells every day on how there's a lot of variance in the cross aunts Andi. That leads to a lot of waste because the questions peak on weekends on to compensate, they sort of create a average amount of cost. Santi Today, which leads to a lot of waste product. We can see how all of these colors have been used in the graph and they much each other, creating quite a smart looking graph quite professional and sort of tells that story using the use of color. Not a lot of multi colors or anything going on in this graph. So that's just a quick way off. How the color palette generated tool is really effective, creating a color palette that'll matches, and then it could be applied to data visualization. Remember, you're just looking for one or two bright colors to be used in your color palette. You also need like a muted gray color on Sometimes you can even pick a matching green and red, so to cover this section that is every way you can use color in your data visualizations to review. Firstly, you can create a clear foreground and background aid in the viewers understanding off the story. Next, you can use bright colors on musical ist. Help focus the attention on the audience to what you want to try and communicate. Thirdly, you can also use scales or conventional colors to add fervent date identity to your visualizations or help come to an understanding of a large data set on Bonnie. Avoid using multi colors on clashing colors, and I would definitely recommend you use a color palette generator, so I'll see you in the next video where we will cover structuring your data, communication and everything you need to communicate with impact, because the graph we've designed here is one that tells a Blair message but doesn't really show you how to put together a full story. We just looked at how to apply Kalitta graphs you'll notice on my little and bakery story that I put together. I've used things such as labels on the graph and text elements to sort of bring out that full communication and story. So in the next section, we're going to understand how you craft the narrative out of your data and how you can then communicate your story with your really nice, easy to understand looking data visualization. So I'll see you that. 19. Analytics Value Chain: hi there. So far we've learned everything we need to know in order to create a beautiful impactful and accurate data visualization. But there's a lot more to inspiring an audience into action than a good looking graph. So without taking a look at everything we've learned already and using it to craft an impactful narrative, then your graph is gonna be all style and no substance. So join me in this section of the costs where we begin to learn how to create a rich and compelling story, effectively communicate with data. So now you've got a well designed, beautiful looking graph. His design lends itself to easy consumption and appropriately focuses the audiences attention on the right parts. But if you want to inspire the audience interaction, you're going to need to create and then communicate a story. So, firstly, how do you identify this story in your data? Many clients I've worked with have asked me to develop an amazing graph for presentation with their data are many occasions when I vast Okay, what is it that you want to get the audience to do or why you showing them this data? The answer, usually in more words is Okay, well, this is the data I have. Onder, when I want to do is get the audience to understand it not particularly effective or doesn't really drive any particular action. This is an example of when data analysis lacks context. Often the context is there. It's just the author hasn't given the proper attention and understanding to the context of the communication, which can lead to presenting data for the sake of presenting data. So in order for us to understand context, let's take a step back and understand Why is it we analyze data? What does it mean to do? Analytics? What is the value Chain off analytics Analytics is often thoughts off as the process of turning data into information, but that only describes part of the process. If you ask yourself ultimately, why is it we do analytics? What is the product of analyzing data is actually taking data and eventually turning it into some kind of beneficial results? The reason we do analytics is ultimately to drive decisions. That's what it means to be data driven is to inform decisions with data decisions that lead to action and results for your business or your process or your activity that you're using analytics to inform if you went into a presentation as a data analyst or someone who's done some analytics work on data you're trying to tell a data story. If you come in and just show information, then it wouldn't be like bringing in the big picture. You wouldn't be telling a full story. It will be as though you've brought in a jig's up. You've got all the pieces laid out on the table in a big heap, and you said Okay, his older pieces. In order to inform Decision now you have to put that story together for yourself, which isn't really the purpose of being an analyst and analyst is someone who takes the data on. Does that process for the audience brings in the communication that then can lead to a decision or and then ultimately into a result. So with that in mind, the full analytics value chain looks something like this data gets converted into information through the exploration and understanding of that data. Then if you apply that information in the right context that leads to incite with that insight, decision makers can then use this data to drive decisions which, ultimately trying to get results for the business or activity or process that they are investigating or exploring with the data. So analytics takes data and turns that into results. So let's look at the analytics value chain by taking a look at a example. Anyone who does any type of analytics at any level from a something really simple, like looking at a table to something extremely complicated like a data science project takes data, and then they seek to understand it. This process off understanding and expiration is the results of analysis. When you do an activity like that, what you're doing is doing some kind of analysis of data on what comes out of analysis of data is basically a bunch of fax or just information. For example, here's a very basic table that shows the sales for one particular product over 10 different stars. It has a list of 10 stars the number of units sold for one of the products on the revenue generated in dollars. This table in itself is just data. It doesn't tell us anything. On the surface, it doesn't mean anything to anyone. It's just a stare. I'll list off data points if we were to study that data and explore it, what we learn from it, or what the product that comes out of it is fat and information. An example of some piece of information or fact from this data is that Star for sells their units $20. We can work that out by dividing the revenue by the number of sales. So Stoffel cells units for $28. That is just a piece of information, and we learn that from looking at the data and understanding it. So store for selling units for $28. It's just the fact we don't know if there's any context behind that. We don't understand if that's fact is a good thing or a bad thing. We don't know what the impact of the business of this information is. Should a decision maker take any action because of this? No, not really. This statement alone doesn't tell us anything, and it can be used in a decision making process. It's just a plain fact that we learn from the analytics. Put yourself in a meeting and you're the manager off this star stone on the far you asked for a part of the sales off this particular product, devoid of any context. Someone presenting, They get up, they've done their analysis, They show you the presentation and they come back and say, Store for sells products or $28. They just present that you wouldn't mean anything to you by this. There's nothing to understand from this. It doesn't provide you any context. It certainly isn't what we call in sight. It's just information. Sure, you understand what star for is you know what the units are. And now you understand what the price of that unit is being sold falling star far, but it's just a fact. So now suppose you're the manager of stuff far again, you're in the same meeting, But this time you're told that Star for is selling units for the cheapest price of all stars, which means that despite being the lowest star in terms of total sales, you're generating the least amount of revenue. Customers are prepared to pay more for this product, as evidenced by the fact that over stars air increase in the price un increasing their revenue from this product. So you should, too. This is an example of insight. You can now make a decision from this insight to then increase the revenue or the price of this product, whereas before you just told context list information. When you view that information in context of over bits of information coming together in the context of the business or the environment or the activity, then it becomes insight on within sight. You can use that to inform decisions in this case start for selling the product for a particular price is just fact. Whereas if we look at it in the context of the over stars, how much revenue is being generated from those over stars? What the price of the product is at the other stars and then look at it through the lens of the store manager who has the authority to be able to make some of these decisions. Then it becomes insight, which led to a decision on then that led to a result. So that is the flow off analytics. When you do analysis on when you're presenting stories, you really need to ensure that you're not just presenting information, you actually have the context and the understanding of the business so that you can present insight to them. So how do we sort of learn how we get to the context of the business so that we can ensure we are delivering insight and not just information, so to come to an understanding of the context? Ideally, when you're giving a piece of analysis to do or your ass to look through some daytime presenter story, you're not given such solid context. The reality is on. In my own experience, people tend to say, Here's some data. Now take a look. Explore that data and tell me what's important of it. That's just the reality of it. They're not going to give you a full list of everything you need to know in order to sort of turn information into insight. They send to say, Look at this data and tell me what it says. Or maybe you've initiated that you've collected some data and now you want to sit down and try and understand what it means. Let's take a look at the same example toe sort of understand how you can then take that initial request and turn it into something insightful by trying to get to the crux off the context of what it is you actually analyzing. So if we take that same sales without any context, you might end up presenting something like this. It does follow. The design principles were laid out earlier. It's created a background in a foreground, uses color appropriately focuses us on the right part of the data visualization store, for it highlights the fact that store forest selling units for 28. It's clean, easy to understand, but it's just information. It ultimately won't lead to driving any action. So if you're not given the full context analysis, you're starting with some data and want to communicate what it means. Then you need to sort of learn the context. Andi. That means being a bit of a consultant sometimes. So let's take a look in the next section and how we're gonna understand the context and really trying get the context out of the requester. Or sometimes, if you are the requester, how you can understand the context of the business so that you can drive insightful data communications when it comes to gaining on the stunning of context for analysis, There's two things that you need to try and learn the who under what the who refers to your audience. Who are you creating this communication far? Andi. When you do this, you need to try and identify different audience members and strike that right balance between two specific, like James is my audience. But also you don't be too broad and just say management. You try and identify different audience member groups who have a different perspective on the data. Stakeholders are too generic and specific people are too specific. Toe. Understand the who think about the different audience groups you're going to be communicating to a new audience group is anyone who would have a specific perspective of the data or the business context. In often case, you may find that you have multiple audiences for one piece of analysis, which may require you to create different versions of your story in order to resonate with these unique audience groups. As a basic example, you're providing a monthly review of your department. The finance team may have a different perspective from the hate John manager. So before you begin any data visualization, just take a step back, look at everything together and try and identify who your audience groups are, then consider if they may have a different perspective of the data and group them on perspectives of data. And then you can work out if you need to create any specific context in your communication for these groups, or if you may need to create separate tailored communications to them. So now this brings us to the what the what? Ask you to consider what it is you want your audience to know or do. What is the outcome of your analysis? Again? You have to be specific here. Things like I want the audience to be updated on the numbers, or I want to inform my audience off the monthly report on specific outcomes. These air too broad and generic. Remember, the value chain of analysis is one that leads to a decision, and an impactful data story prompts decision off specific outcomes. Sometimes what the action might be is driven by the story in the data. You can't always identify the what until you've actually reviewed the data. But before you begin to analyze the data, you can prepare yourself by understanding the context of the analysis. Consider your audience once more and try to understand their business and their perspective of the data. It helps to think of things in terms of levels that can be pulled each lever referring to a potential action. Let's taken example and think of a marketing manager. But they're doing some analysis off different customers and how marketing campaigns have done some potential levers they could pull or potential actions that they could take our things like changing price, changing a market segment, altering the brand, spin up or decrease campaigns. Since you know that potential actions, you can then frame the analysis to lead to some of these outcomes. But unless you understand the potential levers or potential decisions that could come out of a piece of analysis, then you can't really tail your analysis to them. Once you know what the action or decision you want to try and guide with your analysis, you can then communicate with your data to tell a bespoke story. Although I understand that sometimes in reality that just isn't such a solid outcome of a piece of analysis, you may find that there just isn't enough proof yet to make a decision, which is totally fine. Sometimes the point of a data communication or the story you want to tell results. In further analysis, consider a marketing campaign. Once again, you've done some analysis of different customers and identified different segments. The outcome of a presentation like that could be the They want to investigate further on the highlight its segment and try and target potential campaigns to them, which requires additional analysis. In cases like that, the outcome you want to communicate is that you need additional information, and then you want to highlight the date of the you'll need on the potential outcome of doing this fervor analysis. Sometimes that is the reality, and it's totally fine to communicate a story where the outcome is to continue doing analysis. The whole point of the analysts is to get to the decision, and sometimes it can take multiple rounds or back and forth with an audience to reach that point. So we've learned what role context plays an analysis on the value chain of analytics and ultimately, why were telling data stories? What is the outcome of doing this? Why do we do them? But the reality is you are always handed all of the relevant context to you when requested to do any sort of analysis or prepare any sort of communication with data. So how do you actually come to understand that context? That's when you've gotta step into the shoes of a bit of consulting. So join me in the next video, where we discuss consulting for context. 20. Understanding Context: So we understand the important role context plays in a piece of analysis. In an ideal world, when you're asked to do some analysis, the request that gives you a full breakdown of that business, the background information, potential actions they could take or suspicions on context they have on the analysis. But the reality is this is rarely the case. This is when you need to step up and ask certain questions off them. Act is a bit of a consultant to try and understand the context that's needed for the analysis. The reality is the quester usually gives you some data and says, review this data and get back to me with her apart, they have the contact in their head, but they don't often articulate it, or they just assume it's common knowledge or something that you know. It's up to you to consult with them toe, understand the context. There's a lot of different questions you could ask them. I find it best to actually sit in a room with them and sort of go through a list of questions I turned too often. These questions are things such as, Are there any specific questions you want me to ask of the data? What are some potential things you think I might find or should be on the lookout for? What is the background for this analysis? Why do you think it's needed? This is just a few sample questions that you could ask. A full list of these questions can be found in the learning Resource Handout. I recommend you take it with you and whenever you are asked to do some analysis on basket off the audience members or the key people who will be in the meeting or the decision makers so that you can and tailor the communication to fit to their perspective as well as you can totally provide the full insight of the date announces not just the information, even if you think it's somethin routinely you're doing. I use these questions every single time. Even when I'm doing a mostly report off the same data month on month. We have a stakeholder that I know well. I still sit them down and run through the questions because ultimately they know that business best. It's good to eliminate any assumptions from both parties, and these questions is just a great tool to do that. I find that going through them in a dedicated time slot really makes the analysis meaningful for that person. Just having the time and space to think through the analytics turns the Month on month report from something that is done just because it's always been done into a natural, insightful piece of analysis that gets the a proper attention that it deserves. So whether or not it's routine analysis or bespoke analysis, I talked requests for data announces, or just going through the motions of Data Analytics. Somethin extremely complicated. Or you think it's particularly simple piece of analytics work. I recommend each and every time you do a bit of consulting for some of the context with the key audience members. Like I said, it's not just about asking the questions. It's sometimes about giving the audience the proper time to sit and reflect on the analysis and sort of give you the ideal context so that you could make some been really meaningful and impactful that resonates with them. So that's how you do some consulting for analytics. So join me in the next video, where we talk about the types of narrative that you can apply to your analysis on how you can use these general and narratives that resonate with all audience members and use them as templates off frameworks for your design and formatting choices in your data, visualizations and stories. 21. Tales from Work: The Importance of Context: how that so I thought I'd illustrate how important is toe learn contacts with a personal anecdote. So previously in my career I worked for one particular department, and then later on, I moved to a centralized analytics team. At some point in time, the previous department asked me to do a bit of analytics work for them. They handed me some data and said, Josh, can you tell me anything? We need to know that important with this information? And then I thought since I previously worked in its department, then I've got a lot of context already already understand the do's and don't I know how things work. I know what their operations are. I know everything I need in order to analyze this data on present something insightful. So that's what I did. I took the data, I did some analytics work, and then I presented it to the key stakeholders, thinking I'm gonna blow them away. I already have all the context that couldn't be better placed for this piece of analysis on bond. It didn't really have the reaction I expected. It was kind of lukewarm. If I'm honest, they just sort of took in said OK, this is pretty good, but I think we need to look a bit deeper into a few things. So I went away again. I did a bit more analysis and I came back thinking This time I've nailed it, totally made something impactful and amazing, and it's really going to resonate with them on again. I got a really lukewarm reaction about it. So this step repeated, possibly one or two more times where I'd go back, dig a little deeper into what they want. They'd add a little bit more information and then I present something and it didn't really have the impact I thought it did. And then I'm beginning to think, Well, I'm not sure they entirely get what it is I'm trying to say Maybe I need to spruce up my visualizations a little bit. Maybe I need to come at this from a different angle and equally I think they were going back and thinking, OK, this guy doesn't really get what it is we're trying to say here. So this is when I decided Teoh, I need to actually do this consulting for context. So I brought them into a meeting room and we sat on, I went through all the questions that I thought I had the answers to, and they just assumed I had the answers to them as well. Because I worked in this department. I surely know if all the INS announced, I know all of the context acquired, so we took a step back on, began just going through the same questions. What are your assumptions of this piece of work? What do you think I might find who were the key stakeholders, even though I kind of knew all this already or at least assumed I did. And then what we found was a really great collaborative space where we can discuss the different ideas and anecdotes on them Presumptions we each had on this data or this piece of analysis. So after I did that, even though I knew everyone in the department, I knew the key stakeholders. I knew their operations and how things worked. After just running through this questions on the same ordinary piece of analysis, it really brought to light true context for it, which allow me to go back and do the analysis once more. With all this fresh context and they had a lot better faith that the analysis was gonna be impactful. And then I came back and presented it to them, and he was just fantastic. So I hope that little anecdote gives you a bit of information on even though you may, I assume you know all the context already. Perhaps it's a department you already working for, and you're doing analysis in that department. It really does benefit everyone if he run through the consulting for context questions again on every fresh piece of analysis. Because that really helps each and every piece of analysis be tailored and poignant to a directive audience. So I thought I'd just share that with you guys. Thanks for watching. 22. Data Narratives with TEMPLATES : understanding the story that you want to tell. It's just the first step to communicating that story Once you know what it is you want to say. Next you need to choose the type of data narrative you want to structure your story with when I've looked through the data and I know what story I want to communicate. I then consider these different primal story arcs to see which one will best conveyed information that I have. These are the basic arcs that all good stories have founded on, and the same could be applied to communicating with data. You can apply these to your data communications to tell a compelling story, one that's going to resonate with the audience that you have these data narratives types will help your audience better understand your numbers by putting them into context of a story pattern that people readily internalize on. Consistently seek out by having a narrative arc touch to your communication. It helps you frame where the focus and importance of your visualization is. Personally, I think of these almost as frameworks when I'm telling a day two story. I've arrived on the analysis and I know what it is, I want to say I think of which arc this story best fits with. And then I can turn to my sort of templates of stories and helped structure what is I'm communicating and guide me on how I want to set up that communication and format. My visualization. So now let's explore these narratives. Andi, just for you cost followers. I'm going to be sharing you the visualization templates that I've put together that actors my framework, the first and possibly one of the most common story arcs is the underdog. The underdog is a story that every audience knows and one that we all love. This narrative works well in data when you want to communicate a story of overcoming a challenge or inequality, for example, like in this visualization. Hear what I'm doing is showing how a sales campaign has overcome a low seasonal period of sales. With this story arc in mind, I can now apply the formatting, and it helps me frame the story that I want to tell. With this story arc in mind, it helps me apply the formatting to my visualization and helps me select the data that's actually going to tell that story being an underdog story, it's one that overcomes a challenge. The challenge in this case was a low seasonal period of sales, so this helps me highlight where the focus is. So the focus of this type of visualization is one that highlights where that overcome has happened, where the triumph is for that it helps him free my formatting. So I've picked bright green colors using convention toe indicate that this is a tremendous positive thing for the business. I've got my new campaign highlighted in Green. It's signaling to the audience that this was the underdog. The new campaign is what overcame the adversary. I'm highlighted in green on my line graph where that campaign has overcome the sales. Dep. I've placed importance with my indicator seen 60% increase showing off by how much we can celebrate this triumph on. Then everything else of the graph has been placed in less important. So the rest of the graph is muted doled colors because I'm really focusing on the under duck. In this case where we overcame adversity, you can sort of see, I hope how these different narratives help me for a my thinking into designing a graph like this. So let's continue on. We're looking at some of the over story arcs on sharing some examples, and these are the sort of templates I constantly turn to when I'm trying to tell data stories. One of the next essential story, Ox, is one of redemption. Redemption stories convey the theme of hope for the future, applied as a data story. This is good when you want to communicate to the audience how a specific factor could bring about somethin positive in the future. For example, with this graph in this example, one particular thing will lead to a positive outcome. For the business in this case is the size of the market of the South. Region has grown by quite a lot when compared of the over regions, which in the context of this business, if they captured this increased market growth for this section, it could lead to a huge increase in revenue and sales. For them, this is a story of redemption. One positive aspect has come to light that could bring somethin positive in future for this business. Just help me frame my story again. The part of the story that are important here is the particular aspect that is now matured and grown and offers an opportunity for the future. The South region's growth. So what I haven't done has shown you how many sales a generated in the different region or how much revenue is generated in these regions. All I've done is focused on how much thesis ouch Regions market has grown on. Put it in context of the over markets, then have used that as the bright colors drawn in the story on divided certain components. By suggesting, if this business captured all of this increased market, how much does that actually turn into in revenue? Is this sort of framing that data story touching it toe one off the main story arcs, which has led me to sort of create a graph like this. So this is another one of my templates that you can use in your data stories. The next quintessential story arc is one of betrayal, betrayal, being a very common plot point in different soap operas because everyone loves to hate on a good villain in data visualization, the theme of betrayal is when you want to highlight to audience a particular threat to the business. Or you want to focus on the negative impact of the business that some external factor has had a part in. For example, take a look at this graph from my formatting templates. This one shows how a future threat could happen to the business. What we have here is a business that employs a lot of engineers on. There's a future threat that the lack of qualified engineers in the market means that this business may struggle to hire the number of engineers they need to keep afloat. So you can see how the formatting and layout of this graph is told that story. What we have is a lot of Red's going on. There's something negative happening here, so the use of red has been employed to sort of imply that to the business. The specific threat has been highlighted in red. The impact of the business is really what stuns out in this chart. It's not just the threat, but it's also the context of that threat to this business. You can see that in the 13% yearly attrition with losing 13% of our engineers in this business, why that's the case has been explained in this data story. What it doesn't tell you is an over need of information. We're not explaining to the audience how many engineers we have, exactly who they are, what the names are different potential actions that might come of this, which is the result of this communication. The story itself is one that focuses directly on the betrayal on the negative thing that's happening to the business, and it really spells it out as a threat by highlighting parts on highlighting the specific impact that's gonna happen on this business. So this is another one of the templates that you can borrow for your data visualizations. The next quintessential OG is the victory similar to the underdog. You're communicating something positive that has happened to the business. But unlike the underdog, which shows how something overcame, you're really laying out the context in an underdog story off exactly what was overcome and how you overcame it and what it led to in the business. The positive outcome this, um, overcoming hod in the victory, you're just celebrating the success. So in this graph, remember? Sometimes you don't need visualization to tell that story Sometimes a really nice formatted number in context really gets the point across. I find when I'm trying to communicate just an overall success of something, then placing that number in context really gets the victory story are communicated. In this example, we have the complaints rate were above target. How we're above target last year and won Our complaints rate is right now. I've used really bold large numbers for the formatting in Green to celebrate that success, this sort of formatting template all it does is show the success. Show the victory, and that's all that's really needed in this type of data story. So it's nothing overly complicated. It's quite simple, and I find it to be usually the most effective way to just celebrate the success. And the final story. Arc is one of tragedy. Humans, a drunk, the tragedy and in data narratives, tragedy often focuses on something negative that has happened to the business or the context that you're communicating in similar to the previous example. Sometimes the number is all that's needed to resonate a powerful way to communicate. The absolute negative impact of the business of something is to communicate with big bold choices off visualization. So here I've got huge red numbers taking center stage to really let the empty really emphasize the negative result to the business and just let it resonate with the audience. So they considered stew on what has happened here. If you're not really communicating that a threat to the business or you're not, the purpose isn't really to drill into what led to this result. And you just ultimately trying to communicate how ineffective something waas or overall negative impact to the business. Then this story arc is for you. So these are the quintessential story arcs that form data narratives, and I use them as the framework or almost 10 plates in any data story that I tell. So whether or not I'm communicating absolute victory or defeat, I'm trying to highlight a potential opportunity and what that might bring to the business. Or I want to highlight where an event has led to success or failure. Then I turned to one of these data narratives which feeds into how I'm gonna format them how I'm gonna present my communication on overall, how I'm gonna contextualized this story arc to the business. So so thank you for sort of taking part in this learning that I've put together. Then I'm going to share all these with you. You'll find them in the learning resource is of this cost on. I hope they bring you some value to help you frame your data narratives. So remember, once you've got your communication together, you've done the analysis of the data. You know what insight is provided. And you know what? You're trying to safety audience. See if it fits into one of these data narratives and then you can use these frameworks as a template to put together your communication. 23. Turning a Graph into a Story: so the final part of communicating your data on your visualization is to turn it into a story. So far, we've learned how to create a clutter free impactful graph that highlights attention on. We also know how to format to make you look smart and presentable and follow some general design rules. And we also know how to sort of learn where a data narrative comes from and understand all the context necessary to present insight. So finally, let's take a look at how we can take our well presented graph from something that's just smart and cleanly communicates data on how we can apply formatting to really use it as a foundation to tell a story you would have seen from the previous examples. Some additional elements that I've added on to the different visualizations isn't just the data presented in a well formatted way. There's a lot of different things on the graphs like call outs and indicators on where I've used the title or even added just written text that'll complement the narrative, says Lin. What we call these different components and how we sort of apply them to our graphs to really tie everything together and tell a really great story. So once I've got my graph, the first thing I usually change is the title. The biggest pitfall of a good communication is the title when most people create grass, the user title that describes the graph rather than one that informs the viewer off the story. When you present your visualization to your audience, they begin scanning the page for clues about what it is they're being told. The first thing they actively notice or read is often the title, so why make it one that doesn't actually in farm? Let's look up a previous graph we've seen before the title Hair tells a story. After reading the title, the audience now get to clue as to why there's an increase in the trend because of the new sales campaign. If we make the title one that is less informative and simply describes the graph, the audience is non the wiser as to what the story of the data actually is. So it's best practice to avoid creating titles. Simply described the graph. If you look at a botch up that shows you the sales of three or four different months and then the title says sales off 3 to 4 months them What's it really? Adding, It doesn't really communicate the story. So avoid using titles and a simply descriptive off the graph. The title should be one sentence or a few words that summarizes the key aspect of the story , such as in this example that says the new market is what led to a sales increase. As a poster saying sales poem of from January to July, that is one that doesn't really say a lot. A new audience, after looking at the data visualization, typically turns to the title, especially if you've used all the formatting rules that we looked at earlier. So the title should be one that tells them in a quick summary what the story here is. That then allows the data to be reinforcement to that story. When you're setting out to create data communications, what you're doing is basically writing the story on the page, and then the data graph and the visualization itself is the supporting evidence. By having a title that is simply descriptive off the graph, it doesn't tell that story. It doesn't allow the data to be the evidence. It tries to make the day to itself. The story, which doesnt effectively communicate to the audience. So remember, keep your titles informative are not descriptive. So the next element you can introduce your data visualizations our call outs, call outs and the bits of text and labels placed directly on the graph or around it. Sometimes it's an indicator number, a line or arrow pointing to something or text description, sometimes off to the side of the graph. Some graphs lend themselves better toe, having call outs written on them things like bar charts, plots and line graphs, and sometimes just the number itself placed on the page. They give you a lot of white space, which you can use to add labels and right on and help inform the viewer, which really is just another reason I hate pie charts. Pie charts don't give you all this space because the graph itself just takes up. All of the color on the page takes up a lot of space, and you just can't effectively label on them. So again, if you needed another reason to hate pie charts, here's another one. So back to call outs, what are they and what is best to use them. I find there's two ways to best use what I call call outs on the page. Firstly, is to write the entire story. You're trying to communicate as a small text box placed next to the graph. So you'll see from some of my template ID story ox that they have this entire story right now in full sentences for the audience to understanding. Consume. I find this really helps to communicate that story, especially when if you've created a presentation or some kind of report on, then the audience views this medium. Later, they may forget all the conversation they hot in the communication, but they won't forget because it's written down on your graph for next to it, so they can totally understand the full story. It also helps to place it in small, faded, muted text, so it doesn't take a lot of priority in their attention. But once they come to a full understanding of everything, they then turn and read this little bit of text, and then it explains the whole story to them. Next is to label the graph directly. We have little sentences or few words to sort of completely spell out that story. This works especially well when you're doing a line graph over time, so you can see how in this graph I've used both for these to come together and tell a full story. This graph here employees call outs in both of these effective ways, simple text of the right wraps up the whole story that the audience has just been told on the graph elements on the data visualization where have included some text on this dash line. Really tell the audience exactly what happened at this point in time or will happen at this point in time and sort of completely spells out what this graphics saying you'll notice in this graph that the data takes second place. The line itself off this line chart isn't really what's in focus. It's used a supporting evidence, all of the different elements placed around the graph. And then I've summarized that whole thing of a little story to the right. So the final element you need to introduce to your data visualization is what I call indicators. These are the little numbers placed around the visualization. It's something you compare with your graphs that sort of give additional information and context about what the audience is seeing. It helped place that data inside some sort of context, which allows them to easily understand more about what the story is you're trying to say. So take a look at this example from my templates list. The main point of the data is that the market signs of the South region has grown by 45%. But this offer has contextualized that we have the indicator that shows if the business captured that market, they can expect a $1.2 million increase in revenue. So this is what I call an indicator. It's a little number place to the side or around the communication that helps contextualize the numbers that the audience is seeing. So in this case, the story is they grew in market signs, and I've contextualized that by showing how much that is in a revenue dollar amount for this business. This really adds a lot of depth to your visualization and helps when creating a lot of data density so you can communicate a lot more to your audience without needing a lot of different graphs or taking up a lot of space over long communication. You can do so by telling a main story with your data visualization and then adding some contextual indicators to help resonate and let that information sink into your audience. When trying to create indicators to your graph, I think less is more. Try and avoid having too many. I find two, possibly three. Most is all that's a quiet when you're creating them trying. Think of your audience on what the story you're telling means to them. What over information might they want so that they can then take this information and insight you provided into a decision? In this example? It's a story of opportunity. So the impact of the business of the opportunity has been applied by use of an indicator. So that summarizes everything you need to know about creating impactful data visualizations that will resonate with an audience and ultimately drive results. I know over the past few hours everything we've learned has been a bit theory based, but I hope you've taken a lot away and can now go ahead and start creating data visualizations and telling your stories. So in the next section what we're going to do is focus on how we bring all of this together . We spend a few hours now doing a lot of theory based things on color, visual design principles and narrative story arcs, and looking at different sections off how you communicate your data and all these different components. So in the next section, I'm gonna go food steps to creating your visualization from start to finish in an organized way so that when you're telling your next story, you could sort of condense everything you've leant over the costs into a few easy to follow steps. And you know what? Order to tackle your story in. So I'll see you in that section. And then after that, we're going to cover some case studies to give you some real life examples and hopefully give you some inspiration on your stories. Thank you for joining me on. I'll see you in the next video. 24. 6 Steps to Telling a Data Story: so far in the hours of this cause, we've covered a lot of theory. We've learned a lot about formatting and different methods to create impactful, data, visualizations and different narratives that you can tell. I truly hope. Along this journey, you've learned some valuable information. So now in this section, we're going to take everything we've learned and bring it all together in one neat package and run through the steps that you can follow to create clear and compelling stories with data. So now let's run through the six sequential steps to telling a great story with data. The first step to creating your data story is to ask yourself, What is the point of this communication? Why did I do this analysis? You need to decide what it is you're ultimately trying to achieve With this analysis. It's always a good idea to ask questions off the audience or key members of the audience of this analytics work. Even if it's a standard, re occurring piece of analysis, think of your audience and identify the different perspectives different audience members might have on. Then plan to create different communications or tweak the communication you have to suit these audiences. Next, After you found your story, you've identified what it is you want to communicate, see if you can match it. Toe one of the Staples Story arcs. This a guide you in how to structure and focus your communication and input into a lot of formatting decisions. Step number two is to pick the right visualization. You need to pick the right graph for the right job we covered earlier. Different visualizations on what the best suited for on which visualizations to avoid. Remember, never pick the pie charts To pick the best data visualization. You can turn to the Guide on Handout in this cause on that contains a great chart picker tool to help you pick the right visualization for the job. Next, you need to edit to your visualization for clarity, Eliminating or necessary legends, Grid lines, tick marks and colors will clean up the graph and allow you to focus your audiences attention on the main point. Consider each element of the graph and remove it. If it doesn't add to the clarity. A good practice is to remove everything on the graph and then add back what is needed. One by one until your graph can be understood and then stop there. Remember, you don't want to do anything misleading with creating and editing your visualizations, so step for after you've created your clear and easy to understand graph is to then format for impact. Used the appropriate colors to create a graph that could be easily understood on focuses attention in the right places. You want to pick bright colors and use shaded, muted, dull colors to highlight and draw attention to where it's needed. You want to create a background in the foreground to make it easy for your audience. To digest the message, remember to avoid using colors that carry context, such as red and green. You can also use the cola picket tool to create a beautiful color palette for your data visualizations. Next is the format for narrative. Replace your descriptive title with an informative one. You don't want to describe the graph for the title. You want to tell the audience a story directly. Then it's a good time to now. Add some indicators that add additional context and framing to the data. Take a step back from your analysis and think of your audience picture them looking at this data and then try and understand what it might mean to them on what additional information they may want. That would then be a great thing to use as indicators Next. You also need to add to your legends or text to the graph directly, particularly if it's ah, piece of analysis over time, such as a line chart. Apply labels directly to the graph that spell out the story. And finally, the final piece of formatting for the narrative is to write a sentence or two or three and place it near or around the graph. Remember to make this small use. The formatting options are faded muted colors to push this into the background, but completely summarize a story in full written sentences about what it is you're communicating. This not only helps the audience understand in your presentation or your meeting or your apart, but also when they go away and they refer back to the slides or the word document or the pdf. They may not remember everything, but if you summarized it in a couple of sentences, it keeps everyone remembering the exact same story consistently, as opposed to leaving the meeting and then having a bit of a different interpretation on the numbers once again because they weren't there to participate in a conversation. So it is always best practice to just fully writes out the story on or near your graph. Step six is what's next. So finally, your data communication should include some takeaways. Andi, I hate the term, but action items remember the analysis was done to drive a decision or to get to a result, each piece of analysis and data story should ultimately have a point. So conclude your presentation or your report, or however you distributed this communication with some next steps or outcomes. So that's it. That is how to master the art of storytelling with data communication on data. Visualization are nutritive living things that definitely somewhat more of an art than they are a science. I'm always learning and improving, so I suggest you do, too. Just keep practicing, so I leave you with an exciting activity where you can put together your own data story. Using a data set in a scenario put together for this cost, I also will share a case study review from a real life example Why take it all the way from the initial request to understand in the context and the role that it played all the way through to creating and formatting my graph into an impactful data story. So I hope with these you can find some inspiration. And now you're fully equipped to go out there and create amazing stories that resonate with an audience with your impactful data visualization. Thank you for taking part in this cost. If you ever want a second pair of eyes on any data communications you've put together, feel free to reach out to me. And I'll give you some pointers on any of the graphs that you're putting together. Thank you. 25. Thank You: Hi that. So this is the end of the cuss. I hope you've learned something along this journey, and you can start creating your own impactful data visualizations. I just want to say thank you for participating in this learning with me. I had a great time putting it all together. I really enjoyed recreating all these different types of graphs to be used throughout the costs. Some of them good, some of them bad. And I really did truly enjoy putting everything together. So I really thank you for taking part in the costs, and I hope you take something valuable away from it. Remember, designing great visualizations is a bit more of an art than it is a science. And you can't just take away from this cause and immediately start creating a fantastic graphs. I think there's definitely a room of continuous learning on practice to be done. So I recommend you just go away, start practicing and doing anything you can to sort of create great visualizations. If you want a second pair of eyes on any of the grass in Ukraine, then just contact me in this cars and I'll be happy to take a look at them and share some ideas, review or collaborate on any visualization that you're creating. I hope you've enjoyed this cause. And I encourage you to leave a comment on a review all star rating on the car so that over learners confined it on. Learn some of the information that you have. So once again, thank you so much for participating on a hope you've taken something valuable away from this. Thank you. Feel free to reach out to me at any time with any questions you have. 26. CASE STUDY 1 - Overwhelming Monthly Report into a Succinct Data Story : hi there. So in the final part of this costs, which I truly hope you thoroughly enjoyed, we're going to review a few riel case studies on Apply everything that we've learned so far to them. I'll start from the data in the context and take it right through to the final communication. I hope these show you how to apply all the you've learned and give you some inspiration on the communications you're gonna put together in the future. Just as a note, you can find these case studies in the handout, so I recommend you follow along. Plus, you can always refer back to them later in the future or just rewatch any of the videos from the case study. If you're looking for a little reminder on what we've learned, so let's just get straight into it with case study one. So in this case study, we're going to review something that is true to life. This is a riel case that I was asked to support him. Obviously I've removed any personal identifying data. All the data has been made, dummy. Andi, all that personal stuff. It's been removed. So this is something I get us quite a lot to help with the monthly report and try and make it more meaningful. So I thought this would be a great first case study because I think a lot of people have monthly reports on Sometimes you find that that none is impactful as they could be. So this company has seven stars to sell a lot of different products, and each month or apart is put together for management. That summarizes the month on main KP eyes. So let's start by taking a look at the report. You may want to look at this report in your workbook, as it can be a little squashed on the screen. Okay, so this is there a part? You don't have to remember all the individual details of it. Just that they are showing a summary off march and displaying some key KP eyes, such as revenue, number of sales, new products that they sell new customers as well as new members to their loyalty scheme. So this is the monthly report distributed around management and over key members of the business. This is quite a true to life example, but there's a little more going on in this case. The senior manager, who is one of the main audience members. He quite likes this report. He doesn't want to see changes to it, and he insists on the color scheme as it matches the company brand. So our task is to make this more meaningful without making too many changes. The I'm gonna really gel with the senior management, possibly you could think of when this has been applicable in your business. Perhaps you've had a monthly report that you've been asked to look at, and you find that some people just enjoy the way it is. I don't want to see changes or resist any updates to their apart. This is something I've been asked many many times to look out. A monthly report is produced, but it's quite detailed and doesn't contain much narrative. It's just full of too much data, meaning it doesn't really have an impact. But it's made every month on a lot of audience members have just gotten used to it, so they resist any changes to him because they believe doing so might put them at risk on missing out on something they say. Why change what isn't broke? So make sure you're familiar with the report. We have him pause the video if you need Teoh and have a look over it in the Hunt Out book before we get into the review. So let's see what we can do in this case study. So a monthly report like this is extremely common practice in most businesses, on often some audience members or consumers of the report. Perhaps there's a data savvy person in the department. Or perhaps it's you and you identify the A report like this doesn't deliver on as much insight or story as it possibly could. So ah, proposed always made change. The reports on ditz often met by resistance from those who like it. The reason this happens is because a monthly report like this is fantastic at the job it does, and in a way it can't be replaced. A report like this delivers all of the information needed to make decisions. It very succinctly summarises Ah, whole load of data and information on, then presents it in a format that could be understood easily, especially by those with the proper business context, which it's exactly what is the intended purpose of this report a monthly report like this isn't designed to tell a story, and therefore it can't tell a full, complete and poignant story because it isn't intended to. Nor could you cram story full of all this different data that you find in the report and still keep that stories a sink That would be more of a data novel than a data story. So what tends to happen is people who want all of the data presented, like the report on those seeking us a sink story from the numbers don't like it. So when the port is being replaced or changed and a proposal is made to do that, resistance should be expected. With that in mind, you don't actually need to replace their apart. What I always recommend is to supplement the report with a story. Let the report contain all of the numbers on dedicate a page of the report or you compare it with a presentation that tells a complete story. Let the numbers of the report be the index to that story and supplemented with a separate communication that tells the story you want in my own experience. What I found is if this story part of the communication is done well, then, over time you find that management become less reliant on a big report full of numbers, become more interested in the story that they tell, though. In cases like this, we create a separate piece that tells a story on then parrot with the monthly report. However, having said that, it doesn't mean we can't change the report a little bit so that it is a bit more impactful . So before we get into creating a communication that tells a story, let's see what we can do with this month for your part to add a bit more value to it. So for this particular month, the story that the offer ones to tell has two components. One. They want to share the success of the loyalty program, which has had its overall best month in terms of new members since the launch of the program. So this scheme is really picking up momentum. Secondly, Stall Seven is a recent star that they added on opened just a few months back, but it isn't doing as well as expected. It continues to be the worst revenue generating star, which management hoped would increase over the months, but sales aren't picking up the star. Okay, so let's see how we can incorporate thes story elements into our monthly apart without changing too much of the reports or thinning out the data. We want to keep it familiar to the key audience members. Remember, don't think of the monthly report is something to be changed or replaced. It has a place and a job that it does, and it does it well. Think of a monthly apart like this is something to supplement a good data story. Okay, so this is my revised version. You'll notice. On the face of it, it's retained its look and feel. The key members should still find this familiar. But let's take a look at specifically what I've changed and why I made those changes straight away. You should notice the boxes or cards that act as a bit of separation between individual sections of the report. This just makes presenting a lot of information a bit more digestible. For people. Onda as an increase is white space. It gives the report a brighter, clearer kind of feel to it. You should stop at the top and notice the at a glance section, where I just put down a couple of key bullet points that summarizes the theme of the monthly report. Then I want you to notice to the left of the data tables. I've also added a few words on what they shouldn't notice from this data set. These are just small design choices to make the report a bit more digestible by compartmentalizing the sections as well as dedicating a few spaces on the report that summarizes the story and makes for a great scene setter so that we can impair this report. We have a more bespoke communication. It keeps the report doing the job. It's in tenders, a dupe summarizing large data sets but also incorporates the themes of the story without diluting the amount of data the report contains. So the management should be happy. The report has been kept and retains its look and feel on purpose. But now we compare it with an appendix or a deep dive into the story. So let's see how he can put together a bespoke data story to tell the narrative we want to . So as we pull together this data narrative, we're going to be doing so by following the six steps to a Day two story, which you can refer to in the handout books if you want to follow along. But Step One is what's the point, which asks us to take a step back and review the context of the story for the purposes off the case study, we can assume that I have gone and how to speak and had a meeting with the key audience. Members and decision makers on have gained the context, which I got from asking the questions from the suggestions in the handout book to remind ourselves that story we want to communicate is that Stall seven isn't growing as management had hoped. And secondly, the loyalty program is really successful achieving a new monthly high in terms of new members so we know what it is we want to say. So I've gone ahead and had to speak to key members on ask them some questions to learn the context and what I learned. His management assumed that Store seven would grow a lot faster than it already has. They also assume that this is because the market was more challenging than they anticipated , but they believe the store will increase in its revenue and sales soon. Secondly, the new monthly loyalty customers is great. That really pleased with the result. I asked them if they have any suspicions about the lawyers of program in the coming months or what do they attribute this success too? They said. The success is probably because of the great benefits it offers on also how each story is doing a great job with promotion. They don't think of anything in the future. They expect that it will just continue to rise. Okay, so these are important bits of context that we can then use to tailor our communication. We will look at each of the stories individually. So, firstly, the failure of store seven now the next step to creating a narrative asked us to see if any narrative templates fit this story, which will help guide us with formatting and structure. So we know Store seven is not achieving the expected sales, and this is a negative thing. We know that since we explored the context, the management are continuing of efforts install seven on believe it will catch up. So you know we are just trying to show how something negative has happened. What were actually showing is a threat, a potential bad thing in the future. I store seven continuing toe underperform. So we know the story arc here is one of betrayal. We're highlighting West. Something endangers the business. To step two of the process is to decide on the graph. This is an easy one. Since we're showing how something has changed over time, then we're going to pick a line graph. So let's take a look at what that now looks like. Here is just the default line graph our starting point of telling this story. So this is just the data points plotted, have used a line graph tool and created this very basic line graph. I've been labeled it The chart title reads Chart title and it says Siri's one. So we're going to start from here and use this is our base to start creating our story. The first thing we want to do is clear away all of distracting graph elements that dilute the message. This should be a pretty easy step, because this graph was pretty Uncluttered to begin with, but I still made a couple changes. I got rid of the vertical axis and just applied the data labels directly to the line itself . Then I removed the legend on Dive started by making the line read. Since it's a negative, let's reinforce that. So that was a quick and easy step. Now we want to move onto the next one for matting for impact by using the story. Arc template is our structure. We know in this type of story we want to show the future danger to a business. That danger is, if small remains below expectations. So let's incorporate that visually by showing where the expectation is so we can emphasize how far off the mark Star seven actually is. So I've extended the horizontal axis for an additional couple of months, and I've extended the vertical axis and added a sales target line for Star seven. It isn't very pretty right now, but we're going to change that. But we can see our story coming together. Now we have a graph that shows the danger to the business by how far off from the target Star seven actually is, right. So we've got a pretty good graph which contains the elements for the story. Remember, the story in such communications is something that you tell to the audience. The numbers are just here for evidence on they just support the starring. So you might think this graph is a bit sparse right now, but you'll see how it comes together to complement the story elements as we begin to add them now. The next step is we want to focus our formatting, which is everything we learn from visual design, visual perception on some of the do's and dont's of grass. So if you're coming back to this at a future time and you want to recap, you can go back to those sections. At this point, we're just trying to bring out the focus and formatting of the graph. We're gonna add the story elements a bit later. So now I've added some formatting to the basic graph. I've colored both lines, read as thes reinforced the negative nous of the news. But I have distinguished them by using a solid and dashed line which helps create separation and helps the audience identify that these represent two different things. I also indicated a forward looking months with a faint a line. I've also labeled the lines directly, which makes them a bit easier to read. I plotted the graph on a background color, a very muted light one. So this really places things in the background and emphasizes the data elements in the foreground. I also find this helps to create a separation between the graph on the over elements such as the title and indicators that we're gonna add later. I think it compartmentalizes the graph for the audience and makes it easier to understand and makes the narrative pop our A bit more so I recommend. Generally you can use background colors as your background to the graph itself. Those of you with a keen eye may also have noted that I removed the data labels on decided after all that they're probably going to be better on the access, which is okay to do in designing your graphs and formatting them and telling your stories you might find you go back and forth between design choices until you get something that you're pleased with, which is totally fine and something I still do all the time. Now we're ready to add some narrative elements to our graph, which we also need to then fall Matt for impact as well. I like to break the narrative into two parts. Call outs and indicators, then, is the story, which means the whole story in title. So let's begin move on call outs and indicators. You'll know that this story shows a future outcome or danger, so we can contextualize that by showcasing the business result of this danger as well as contextualizing just how far Star seven Co only is from its target. So I'm gonna put those values in as indicators on we can also are to call out to the graph to explain it a bit better so you can take a look of those changes here. Now we have a nice looking graph. We have timelines. It's always a good idea toe label some point along the time line, indicating an event that impacted the graph so you can see that's what I've added with the line that reads Store seven. Growth slowed undone the right. You'll see some values that contextualize the threat. The current state of the threat being 3.4 million below target for the current month as well, is if you add up all of the past and future mums and work out the total difference from Target, then that's what I've added where it reads 12.2 million. So now the final aspect is to just add the story to the graph, which is done by literally in writing the story next to it, as well as a solid title, one that doesn't describe the child but explains it. So here is the final result. You can see the entire story has been written in a couple of sentences to the right on the title gives a quick summary in just a few words. So the final step is to consider what next Remember the Value chain of analytics is to convert data into action. So if you're not including an outcome or action items to any sort of communication or data story in business, then what was the point? So what do we want to do in this communication? Sometimes you can't have an outcome unless you fully understand context, which we know. The management are hopeful the revenue will grow and exceed target for stall seven, so there isn't a specific action to take on in terms off correction to the Star. But given that the results are not expected, the outcome hair is possibly to investigate. Why that might be Remember in analytics, on data stories. Sometimes it's a bit of an iterated process. If you find you don't have a specific outcome, then you shouldn't consider the story fully told. You should think of this as Step one in a process to get the action or the outcome. So here the outcome is further analysis on what factors played into Star seven failing to achieve expected results. So we might conclude this presentation with recommendations on specific data points we need , or to spark discussion with the audience on what they think of the reasons which impacted the store negatively, which we can then use as context in the next stage of this analysis, where we can repeat the story and find some new information. So the next aspect of this star we want to tell this month is the success of the loyalty program, which, if we look to us story arc templates on the context that we learn, we know that this is just sharing the success in this case, exploring the context saved us a lot of time since we found out that there wasn't really any need no interest from the audience in exploring the data that just pleased with the result. So I won't go into step by step on creating anything here because it's pretty straightforward. So he, you know, is ah graph. As you know, a simple success. Sorry can be sung in just a few numbers. We don't always require anything more big, bold numbers, emphasizing results on employing Green to emphasize the success. That's all this acquired in some cases. So what we did was we updated the report to incorporate some elements of storytelling and changes. Design a little to make it a bit more easier to read without changing its identity on, we decided to pare the monthly meeting where this report is distributed. We have a presentation and a couple slides that tell a data story for this particular month . The result of this waas great success. Everyone really enjoyed the storytelling elements on appreciated the familiarity of having all of the numbers available to them in the report. What happened after a few months is we found that the senior management began relying more and more on the data story and decided to just replace the monthly meeting with the story itself on. Then the report was just distributed over email for anyone interested in diving a little deeper into the numbers. So the emphasis of these monthly catch ups became. The story that the data was telling are not just presenting data for the sake of it, but that was still retained because sometimes it is valuable to have these monthly reports and share data between the different apartments as they make for excellent and quick reference guides. So that was everything we created really impactful data narrative that inspired some action from this department. 27. CASE STUDY 2 - Conflicting Data into Compelling Direct Narrative: hi there I would like to do is share with you and never case study in this case study. We have a large organization with a dedicated recruitment team who hire a lot of people to promote the organization as a brand toe work. For they pay for advertisement campaigns to post on job boards, websites, social media in different places online. They're constantly reviewing the campaigns for their effectiveness. After a campaign has run for a couple weeks to a couple months and beyond. A media company that puts the campaign together and does all the posting shares the data with the recruitment team on. They share a report as well. That shows how the campaigns performed on they also give the data set. In this case study, the organization run three new campaigns for a total of 10 weeks. They wanted to review each of the campaigns and select the best one to keep running for additional weeks. So they wanted to end two of the campaigns and then keep and continue to run with the best one. After they got the report from the media company, it had a clear recommendation book the full day to set they got made them second guess what the report was saying. So they turned to me to get to the bottom of it and to recommend to them the best campaign . So let's begin by taking a look at the report that the recruitment team received from the media company. So just upfront, you should be able to pick out some obvious formatting Don't since, such as the pie charts on the red and green that don't indicate if things are good about ineffective titles and so on. But as hard as it might be, we're gonna look past that for now. What does this report tell us? To give you a little background, the crew minty measure the campaigns on the number of people who saw an advert, the number of people who clicked on the advert on then the number of people who then submitted an application for a job. At the top of the report, we have two pie charts showing you the clicks for each of the three campaigns on the views for the each of the three campaigns. Then underneath we have the number of views of each campaign weekly across the 10 weeks that the campaign run. Then underneath that on the left we have the cost of each of the campaigns divided by the clicks to show us the cost per click measure. And finally, we have the total cost of each of the campaigns. Okay, so we see on the pie charts. That campaign, too, has the majority of clicks and views. We also see that Campaign three has the least clicks under the least views, far less than campaign number two on the timeline on the line graph. What we can see is the same trend every week. Campaign, too, has way more views than any of the other campaigns. Then on the cost we can see that campaign to is the mid priced campaign, but the cheapest in terms of cost per click. So despite it being the mid in terms of cost for every click it generates actually works out to be the cheapest cost per click. We can also see that Campaign three is way more expensive, both in terms of cost per click and in terms of total cost. So there's quite a clear recommendation. The best performing campaign is clearly campaign number two. It's also the cheapest, and it brings in the most clicks, therefore the cheapest cost per click. But this case study Wasat the recruitment team, unsure if this is accurate because the full day to set that comes with their apart is hinting that perhaps campaign to isn't that effective after all, So they want me to get to the bottom of it. So, firstly, let's take a look at what that full day to set looks like and see why is indicating that campaign to isn't all that effective? So what you're looking at now is the data set that gets sent along with the report that the media company sends to the recruitment team. We can see we have three campaigns. Then we have the 10 weeks that the campaign run on for each one. We can see the total views, the total clicks on the total applications underneath that we have the cost on the cost per click measure. So why recruitment teams saw the previous report and thought campaign to was the obvious choice. It was the best performing on. It was also the cheapest, but they're looking at this data set, and they're beginning to doubt if that is actually true. They picked up on a couple details here that make them suspect that campaign to isn't the best one. After all, if you look at the clicks for Campaign one compared with those for campaign to campaign one isn't that far off. But it also has a lot less views, meaning Campaign one is probably getting a better view to click ratio. But does that mean it's more effective? They also noted that Campaign three has similar, if not slightly more applications than Campaign one, and definitely more than campaign to. So with this conflicting information, how does it all come together? What's the real story? That's when they turned to me. So how did I get to the bottom of this? As with any type of data story, the key to the story being successful isn't in the formatting, although that is pretty important. But what's really important is understanding the context. A good data story is one that hits the nail on the head with the audience. It's one that resonates with their perspective and business context. So to truly understand which campaign should be continued on which to should be dropped, I have to understand the context of what it is that determines a successful campaign. So I spoke with the key decision makers. They said that campaign to seems impressive. It's the cheapest and brings in the most views. I asked. OK, what is the main KP I for this business? This department? What determines the success of this team in this department? They said it's the number of quality candidates that they hire. Great. So how does the views on a campaign contribute to that? They responded with, Well, more views means more successful candidates. I said Great. Okay them. So if you reported the senior management that over this period of time, let's say we had a 1,000,000 views on our campaigns, would that satisfy the KP? I They said, Well, no, because the senior management wouldn't want an answer like that. They'd want to know how many people we actually hired from that number. So I said, Well, okay, then, So views isn't actually a measure of success in these campaigns then. More views means more clicks and more clicks means more applicants, so therefore, it's actually applicants that determine the success. It doesn't matter how many views each campaign talk to get their applicants of the measure of success. So with that little example of how the meeting went on, how we got to the context, we can see that with the right context, we can come to something more meaningful with the data. As it turns out, the applicants are the success factor and not the views or clicks. So the reports sent by the media company shows a perspective of views and not applicants, which means it isn't very effective. So if we look back to the data set, we can see that Campaign three, despite being the lowest in terms of using clicks, actually brings in the most applicants. So what does that data look like from the perspective of applicants are not the perspective of views like it wasin of the original report. If we take a look back at the original report sent by the media company Onley this time, I've made a few changes to it so that it shone from the perspective of applicants. Now we get a very different story. We can see that Campaign three generates the most applicants and in fact, campaign to generates the least along the bottom we can see that week on week. Campaign three performs the best, and if we take a look to the top right, we see it's also the most expensive. But it's actually the cheapest when you consider the cost per applicant and campaign to turns out to be the most expensive when you compare the cost per applicant. So the total opposite story is true. Campaign three is actually the best whilst campaign to is the worst. So context plays such a vital role in telling data stories. Unless you know what the context of the numbers and business is that no matter how good your grass look, they don't mean anything you can imagine. Even if the media company produced some amazing looking reports, they got rid of those pie charts, and they used a lot of the principles laid out in this costs. It would still miss the mark. The media company produced her apart from the perspective of clicks and views, which told one story. But that isn't the right context for this particular use case. They're not concerned with how many views their adverts get, nor they concern with clicks what they actually want to measure is applicants. So there was nothing untruthful about the report that the media company produced. It was this that they didn't have the context. Therefore, they presented factual numbers that didn't resonate with anyone. They didn't truly align to the KP eyes off the recruitment team. Therefore, they don't have any kind of impact. So we've all that in mind. How do we now take this report and turn it into a story? We know we have the right context, and we know what it is that we want to say. Now we have an idea of what we're trying to say. Let's pick the right data narrative template. What we have here is success. So perhaps the absolute success story narrative template is best because we just want to present the metric Well, that doesn't really get to the bottom of it here. If we just presented and said Campaign three, then the recruitment team, because we understand the context that they're actually a bit confused between the campaigns, Then if we just outright came and said pick campaign three, they think, Well, what is that based on? So what we actually are doing here is a comparison. What we want to show is how Campaign three compared to the over campaigns on why we should pick it. Therefore, this is more of an underdog story narrative. If you recall the underdog story narrative template that I'm showing you here, it's one that showing us how this latest campaign helped overcome a slump in sales. On the narrative template has some numbers which contextualized that campaign and that overcoming of the Sales Dept. This would be perfect for our story. The only thing is, we're not comparing a timeline. We're not showing sales or campaign clicks or campaign applicants over time, like in the example of this story narrative template. What we're doing is comparing one campaign to another, so a line graph is not gonna work. So when it comes to these templates, you don't have to copy exactly the formatting. The point of the template is the structure and frame what it is you're doing. It's to get to the core of what the template is doing and how it's presenting it. Not to focus too much on the specifics off what the template is showing you in terms of graph elements. So the core of this campaign or the core of this story narrative template is one of comparison and showing how something compares better to over things. What it's showing us is how one present period of time, where the campaign sales the campaign existed, compares to the previous period in time where the campaign didn't exist. So we can actually use this comparison in our example and our case study on Lee. We're not comparing one point of time to another. We're comparing one campaign to the over campaigns. So what exactly is it we're comparing with these campaigns? We know Campaign three is the best previously, the Media companies center part that compared clicks and compared views. But we know that isn't what separates Campaign three from the rest. We understand from the context that we discovered that the key k p I. That this team measures is one of how many applicants they can get. So that's where Campaign three separates. Therefore, we're going to use the number of applicants as our comparative measure in this data narrative template. And since it's not gonna work as a time line graph, we're just gonna pick the plain and simple bar graph the media apart percent by the company actually already contained this comparative you off the three campaigns measured and compared against each other only they picked a pie chart which, as we know we should never use. So I'm gonna go ahead and create the bar chart compares the three campaigns against the number of applicants they generate. So here is the standard bar chart. As you well know, by now, the first thing we want to do is clear away the destructing graph elements. This one is actually pretty clear already, so this will be a pretty quick step. So here's the revised version. All I've done is removed the Axis labels on the legend. Now let's create that initial round of formatting. Our aim here is to format the graph so it creates a foreground and background and could be easily understood on brings in a bit of focus on the elements we want to highlight from the graph. So here is the formatting off the graph. I've created a separation between the graph on the outside elements for which now there is just a title. I've done this by creating a neutral, muted color for the background off the data graph itself. Secondly, I also labeled the graph bars. I used a bright color on the focus bar for Campaign three under muted, less noticeable, colorfully over bars. I've also made the labels for these bars smaller. You should also know I called them in red to reinforce the idea that that but enough. A little trick, I want to point out to you is that I've put the label for the Focus bar graph for campaign number three on the right of the bar and the labels for the over two on the left. This forces the I toe look far along the access to see the focus bar on. Then the eye moves to the left to see the other bars. This creates a sense of separation between them, and it also subtly indicates that the bath, a Campaign three, is a lot bigger. This sort of creates the sense off moving across a physical space as your eyes have to look from one end of the elements to the over end. It's just a little subtle trick that you can employ with your bar charts. Okay, with the formatting out of the way, let's start by bringing in our labels call outs and indicates is in this story arc narrative template. We use indicators, toe add context to how, exactly? Well, campaign threes performing. We learned from discovering the context of this analysis that the cost per applicant is quite important. So let's at that to our graph in the form of indicators. It's at some context and how good Campaign three is actually performing. So here you can see the indicators for the cost per applicant have now been added to the graph. You should note that some formatting has been applied to the indicators Reds for negatives . I've also used size to end positioning by putting Campaign three on top of the others. Toe emphasize that Campaign three is the one that is performing best. So now our chart is telling a good story. Finally, who needs add some story elements, which is the title on a couple sentences summarizing it as well as the outcome and recommendation for this analysis. So here I present the final data story. The purpose of this analysis is to decide which campaign to continue, So I decided to just use that as the title Remember, Title should be informative and not simply described the graph. I've also placed on the graph to the right one or two sentences that just summarize up the story. So now we have a great data story on. We succinctly summarize why Campaign three should be continued because we understood the context. We can make a graph and therefore data story that resonates with the audience. It leaves an impact in this drive action. In this example, the recruitment team made the decision to continue on with Campaign three. They also aligned their metrics and perspective with the media company so future a parts where, from the perspective of the applicants. 28. Project Submission: hi there said. Now that you've learned everything that you need to know to tell amazing stories, it's time to head back to the project that you did or have been working on and submit both your improved on before versions. If you need to remind on the brief, you can head back to that video and give it a rewatch. In the next video, I will share with you my approach to the project so that you can compare it to Rome. But if your post your project of the class, I will review it and provide personal feedback, the same goes for any data story you're telling outside of this cost, any questions you have direct them to me and be happy to answer them. Or if you want me to review your data story and give you a second pair of eyes, I'd love to do that. And finally, if you liked what you've learned so far, I encourage you to leave a review on some feedback. Follow me on skill share. Why post great examples of projects and data stories as well as personal data stories of I tell with step by step guides to give you Inspiration. Okay, So complete your project and then submit it and move to the next video to see my submission . 29. Project Answer: Hi there. So I'm gonna share with you the answer to the activity. So if you recall we had data for three cinemas on, we were asked by management to review the concession stands of each of the cinemas on answer the question. What was the best performing cinema in terms of its concession stand? So I've put together a few slides. I'm gonna walk you through the answer now. So what we see here are some of the high level details off the data that we had on the left . We see we have the revenue per cinema. From this, we can see that Cinema one generated the least amount of revenue, then cinema to generated a little bit more so a bit more than Cinema one. But then cinema three generated a lot more revenue. This was from the concessions done revenue on the right, we can see that Cinema one sold the second least amount of tickets then below that, we have cinema tube generating the least amount of tickets on Cinema three once again generating the most amount of tickets. But we have to think what exactly? Waas the measure of success. So if we think to the context. If you read through the context meeting notes provided in the activity, you should discover that the actual measure for success is the revenue per ticket. This is because the cinema is trying to maximize the amount of revenue it makes from the concession stand for each of the customers that comes in. There are job and main objective is to generate as much ancillary revenue per customers possible. This is because the profit margins on the tickets extremely thin. But what the cinema can manage is the revenue per ticket. So that is the true measure of what the best performing concession stunned looks like. So if you take a look at this graph on the right, it tells a little bit of a different story. We can see that Cinema one and Cinema three generate almost the same amount of revenue per customer, which means every customer that comes in on average, they generate $4.82 off $4.91 where cinema to generates a whopping six point $76 per customer. So cinema to is actually the best performing cinema in terms of its concession stand because the true objective and true measure is to measure how much revenue it generates per customer. So what exactly is it that makes cinema to stand out? Which is what if you look through the context? This is also what the manager wanted to know. They were trying to get to the bottom of. What could they change in the cinemas to replicate the success of the most successful one? So if you look at the prices of each of the different cinemas cinema to charges the highest prices, the cost for all of their items added together is $10.55 where Cinema one is $9.90 and Cinema three is $9.64. So there's probably a strong correlation between the amount of the amount they charge each product and the amount of revenue they generate. Per person, it doesn't look like a lot right now the difference between $9.90 in $10. 55. But we can see from the previous graph that adds up to quite a lot. So this is our plan. We've understood the context and had a look through the data and you should have identified that the true measure is the cost per ticket or the revenue generated per ticket. And the reason is because cinema to charges more for its items. So this is our plan. We want to demonstrate to management that cinema to is performing the best. You want to highlight the reason of that, which is because the prices were higher on then we want to recommend to them increasing the prices of the other cinemas because every great data story should have some kind of outcome , purpose or action. So this is thesis lied that I put together to tell that story. So you can imagine I've gone back to the management of the cinema and I've brought along a shot Power point presentation, and I'm showing them this slide that tells a story. So let's digest this and break it down to see what elements of it contribute to that story and tell that data narrative so right in the top left, you should notice a bowl title that reads, Increase prices in Cinemas one and three. So I'm really just driving to the straight to the heart of it. This is both my outcome under my title to increase the prices. Underneath that, there's a bit of context because customers are prepared to pay more. Then on the left, we answer this question that we started with, which is the best performing. So I'm showing you that cinema to is the best performing cinema of the other ones. And I've used the cost per ticket metric to demonstrate that I've also given it in a good title that shows cinema to generated the highest revenue per ticket. I've contextualized that by showing that it's $1.85 above the over ones. There's a short little story to the right saying cinema to has the highest prices and therefore generates the highest revenue per ticket. Underneath that, you can see the cost table. So I've just replicated it in my PowerPoint slide, added a bit of formatting to highlight the higher costs off cinema number two and then on the right. I've added some context. This is a story where we want to communicate a change, the business on contextualized it by showing what that change could potentially bring to the business. How does it align to their KP eyes that KP eyes out of maximize revenue for the customers. So here I've done a little calculation where I'm showing the revenue over the last three months for cinemas one and three, and then I'm showing them the comparison to what that revenue would be if they change the prices of their products. So Cinema one changed the price is too much of those of cinema to. It should make around 592,000 on the same for Cinema three. If it replicates at the prices it should make about 812,000 which contextualized is the title. So that's an increase of $393,000 over three months if the prices were replicated. So this is my data story, the presentation slide that I would bring back to the meeting to answer the question we started with which cinema is doing the best in terms of performance of its concession. Stunned. So what we did was we understood what true performance means. What It's a true measure off the performance we understood which one of them have aligned to that which turns out to be the cost per revenue per ticket. We contextualize that and demonstrated what the changes were of cinema to and why it was different to the over ones so its revenue could be increased. And then we contextualized it by showing if we replicated this strategy in Cinema one and Cinema three. What that might look like. So your answer will have cost be different in terms of the look and feel. But I hope you applied some of the same formatting I've done. But the correct answer is one where you got to the bottom of the right measure to show you didn't just show the total revenue generated and across the cinemas, because that would just be cinema three, which isn't really the to measure of success here. Your presentation should also include some kind of outcome recommendation or action item. And you should also be able to demonstrate why that cinema the one you decided, um which should be cinema to, was different to the others on how it contributed to the success. Of course, your formatting choices will probably be different to mine. You may have chosen a different form of communication, but if you highlighted why cinema to Stan's? Oh, you've nailed why the cinema to was considered the best because it had generated the highest revenue per ticket and you added some context to that, then you did a fantastic job.