Transcripts
1. Introduction: My name is Paul
and I'm working as a data analyst for one of the top ten most valuable
brands in the world. In this course, I will teach you how to tell a story with data. You will learn how to address the audience, build a story, support the story with
impactful visuals, and to tie everything
together in PowerPoint. You probably ask
yourself why you should learn storytelling
with data in the first place while working as a data analyst
for a huge brand, I often wondered if
people got my data and my message with it is
action being taken. Are we making the correct
conclusions about our data? It was only after learning
about storytelling with data when I could say
yes to those questions. I think you sometimes have the same worries I
had with this course. You will be able to
improve your skill set and be a more involved
member in your team. And maybe you finally get
a promotion you want it. Either way, this skill will make your work more meaningful
and satisfying. This is the course outline. First, we talk
about the audience. What role do they play when
we're creating our story? Then we talk about
constructing a story. How do we create
a compelling plot and with what structure? Then we go to the visuals, what visuals you should use, and how you modify
them to your needs. The first three
chapters are about the three main pillars
of storytelling. We continue on to
the things to avoid, what biases we have to be aware of and what to watch out for. And lastly, we finish off
with the chapter combining. We bring everything together in a slideshow and
built a final story. You see we have a
pretty tight schedule, but these sections are all necessary to widen
your skill set. You can also use parts of the course when you
build dashboards or other reports where a
fun story isn't needed. This course will be
valuable to you either way. So your hope you stick with it. Download the extra
project file to put all the lessons
into practice. You can try to incorporate the learnings in your
datasets as well. Again, welcome to the course, and let's start right away
with the first section. The audience.
2. Target the Audience Correctly: Lesson one, the audience. Before you start
constructing a story, you should check who
your audience is. It will change what and how
you tell them your story. Ask yourself the
following questions. Are they people from within
the company or external? What are the roads
within the team? How much do they know now? What should they
know afterwards? Are there any
visualizations that have become standard practice? Your story should expand their knowledge,
not reaffirm it. The type of audience can also be receptive to vastly
different style visuals. Let's look at some examples. Example one and example two. This is what I
call a TEDx slide, because at a TED conference you could see a slide like this. It's a very simple one
showing only one number. During these conferences,
the audience doesn't know the subject of the
presentation at all. They need to be introduced
from the beginning and showing them one fact on one slide,
it's easier for them. The most successful
TED Talks have only a few texts slides in them. The slides are only an
assistance to desktop. Lots of talking is done to get the story across at the end, that curiosity for the
subject should be heightened. Now for the other example, when addressing people
from within the company in accompany the
audience is already aware of the inner
workings of the business. You would need more
proof and show more numbers to get
your news story across. In this heat map, we
see a breakdown for each hemisphere and
historical progress. We can highlight important story beats to further
help the audience. The audience leaves with detailed knowledge that
can lead into action. You see the two vastly
different visuals of a different causes. One introduces the audience to a new topic and increases
their interests. And the other is directed
to people already familiar to the subject
and gives details. Both of them can be valid. It just depends on the audience. After you know, the audience, think about what
they know currently. What is the current status
and where our gaps, either in knowledge
or an action, then think about
what they should know or do after your story. If the story does not include the big learnings or
advisors future action, we are only wasting
our audiences time. Think about how your story
will get them to act. The story should be the
vehicle to our destination. To and watch your tone. While you may be the expert on the data and can show
finding spec by numbers. Not everything in the real-world
is reflected in numbers. There are industry properties, practices, and sometimes laws. They can compromise 100%
execution on the data. And knowledge gap on your
side can also exist. That is why choosing an
appropriate tone is crucial. Don't be arrogant, but also be assertive and confident
about your story. Tell them what you
need to tell them.
3. Craft a Story With a Three-Part Structure: Listen to this story. Now that we know who
we're talking to, we have to determine
how we talk to them. And we do that by
telling a story. Ideally, we want to
tell a story with a clear structure to lead the audience through
the narrative. Stories are easier to remember
and to learn from them. Imagine data being a map that
guides us to the treasurer. By telling a cohesive story, your audience will
know how to read them up and remember it. A typical outline is the classic three-act
structure with a setup, confrontation,
and resolution. In the setup, we
establish the context of the story and how the
situation has changed. The middle part revolves around our hero and how the
protagonist overcomes the struggles and
devolves into resolution. The conflict is resolved and the audience
learns something. For a data story, we can use the same structure. First, the setup, formulate
the status quo by describing the current situation and why the audience
should care. Then show what changed by
introducing the problem. In the confrontation part, we expand the problem and
explain it in more detail. We then formulate and guide the audience
through a solution. Explain the benefits of the solution by appealing
to their motivation. And lastly, in the resolution, summarize the solution or learning and
formulate an action. It is a to-do for the
audience or a must know. You see how the middle
part is the largest. This is where we dive deep into the problem and
devise a solution. However, part sizes
can vary greatly. If you are talking about
the day-to-day business, the audience does not
need an extensive set apart because they already
know the current status. But if you're
working on a project and present in front of an audience that doesn't know the details they setup
can take up more time. So ask yourself, how much
does the audience know? If the presentation is based on a project or
day-to-day business. If the action is already done, like in a project, a general structure
could look like this. This is a project-based
structure. You see counteraction
written in red. Describe the counter
action taken for the problem and describe more iterations
if there are any, then present the
findings and evaluate if the measures should be continued or if there is a
need for a new one. Let me walk you through
a story example with a three-act structure. You are a data analyst
presenting in front of internal stakeholders
of a company. They know the day-to-day
business very well. The company struggles
in the market for years now and it's
looking for a solution. Dear. All we know that we are struggling right
now in the market. Our performance is
not as expected. Let's look at our performance
by the three regions. To get a closer look. In region a, we've
performed well with around 50 per
cent market share. In region B. It looks worse, but we are on an uptrend
for four years now. In region C, However, our performance is
continuously declining. This is especially
concerning when you look at the importance of each
region or the total market. We see that region a and B have lost in importance
over the years, where regency grew from
40 per cent in year one to 51% in year five. Regency is now half
of the total market. Another way to look
at the regions is to look at the adoption
rate of technology x. In region a, our
strongest technology x is not that important. In region B, the
adoption rate is larger, but it's getting
smaller over time. In region C, technology x has been adopted
at a rapid pace, making up 55% in Year five. To put the adoption
rate in one chart, we see big discrepancies
across the regions. Now, if region C is so important and technology
x in it as well, how much does the combination of the two makeup of
the total market? We see that it was 28
per cent in year five. This means this segment
is nearly 30 per cent of the total market and should be targeted not only to
improve in region C, but also in the total market. Therefore, I suggest to
develop more products with technology x and distribute them with a strong focus in regency, our forecasted performance looks much better if we do that. Thank you for your attention. This was the story example. Of course it was quite simple, but you get the idea
of a story structure. We first identify the problem, expanded on it, and then
devise the solution. The audience clearly
knows where to go. The story was concise
and to the point. In larger stories, you
would have to debate which information is really
necessary to the narrative. It is more difficult to
reduce them to expand. Another big question
is how to handle figures that don't
support your story. Be careful not to omit any details that don't
support your story. The audience might catch it. More on debt will come
in a future lesson. Also, don't play any games
with visual manipulation. We will look at creating
clean shots later, but keep in mind that pulling visual tricks will greatly
hurt your credibility.
4. When to Use Which Visual: Lesson three, the design. Now that we covered what and how we want to tell our story, we have to figure out how we want to show it to the audience. Visuals play a vital
role in getting you a story to the
minds of the audience. We covered the narrative, but another key part to engage the audience is
by using visuals. Tying your story with
strong visuals will make it much more
enjoyable for everyone. When implementing visuals, it's vital to keep these
three pillars in mind. Choosing the correct visuals, reducing any noise
that detracts from the core message and to guide the viewer to the
relevant information. In this chapter, we are
in the leftmost section. Before you think of
any visuals yourself. It's best to use
visuals that are established and well-known
within the company. The audience is comfortable when they see a familiar design. If they aren't any
established templates, ask for the corporate
theme or any colors that are always associated
with a specific item, such as one competitor is always spread while the
order is always blue. That way the audience has to
think less and immediately knows how to read the chart
throughout your visuals. Keep the colors consistent. Now, let's find out when
to use which chart types, starting with the line chart. Line charts are
well-known and used if you have continuous
data on the x-axis, such as time, you
can use it with one series are
multiple, as seen here. Bar chart. Bar charts can be used in both continuous and
categorical data. On the left, it's used
for continuous data, just like the line
showed previously. And on the right, it's
used for categorical data. You see how the bars are
stacked on top of each other. This is called a
stacked bar chart. To be precise, it's a
100 per cent stacked bar chart because each bar
sticks up to 100 per cent. So there are different
types of bar charts. Charts that have
only one series, bars that are stacked
on top of each other. And a type where the
bars are side-by-side. This is called a
clustered bar chart. However, this type
is really fantastic and you are better off
using a line chart, a bar chart, where
this line charts, you might wonder when
to use a bar chart or line chart when both
have continuous data. If the data is part of something bigger than you could
use a stacked bar chart. Here, the two technologies
together make up 100 per cent of
the total market. By using a stacked bar chart, we see how important X is in
regards to the total market. A line chart, on the other hand, shows how X is more
important than y. What you emphasize
depends on your story. Stacked bar chart
has two sub-types. The 100 per cent stacked
bar chart you saw previously and the
stacked bar chart. The difference between the
types is demonstrated here. You see on the left, the
emphasis is on how much Region C makes up of the
total market in percentage. In the stack charts
on the right, we see that the overall
market is growing and Regency is growing more
than other regions. From this chart, we learned that the total market is growing. This information would
be invisible to us. In the left chart. Regency could have
been growing in percentage because
all regions are decreasing in size and C is not decreasing as
much as the others. But we know now
that this is wrong. However, in the right chart, we wouldn't see exactly
how important region C is. We can eyeball it,
but not exactly. The more data series you have, the more difficult it is to see. You see how similar charts can contain different
information. Keep this in mind when
constructing a story. Horizontal bar chart. A horizontal bar chart is a
bar chart 90 degrees rotated. The usefulness comes in
the flipped data labels. The text is more
readable since it follows the natural
reading direction. In a stacked chart, it
can improve readability. In our region chart, we see more clearly
how Region a is decreasing over the years
while C is getting bigger. A survey could also be shown
in this type of graph, we've replaced the
survey question. The next example is
useful for any type of rankings such as
best-selling items. We see the entry
product at the top. We expand on this. You could also make
this a stacked chart that shows from where the
cells are coming from. From region a, B, or C. Area charts. Area charts are not often very useful as they are a mix
between line and bar charts. In most cases, you could
use either of them. You could use a stacked
area chart to show how technology adoption
evolves over time. You could show it with
a line chart as well, but the earlier chart
looks more impressive. Heatmap. We saw a heatmap in
the first lesson. It is used to visually highlight large amounts of
data in one view. The colors group values together and any
outliers standard out. In this example, we see
quickly where and when the garbage reduction was
the most slope chart. When we look back at this chart, we see the adoption rate throughout the
years, one to five. But maybe we don't need to
change over the time that detailed and only need the adoption rate from
year one to five, then we can use the slope chart. This illustration
shows the key message even more clearly
than the line chart. A slope chart is like
a line chart but used to show the difference
between two data points. Scatter plot. A scatter
plot is used to plot two numerical data series and see if there's
any correlation.
5. Refine Charts: Reduce Chart Noise: Lesson Four, Reduce Noise. Our audience must be able to see the key message in
every visually. This means eliminating any noise that might be distracting. Let's look at common
noise in your charts. If you use Excel pivot charts, you have this gray box
is called field boxes. You can get rid of them by
disabling them in the ribbon. Then we move grid lines
by clicking on them. We don't need them
because we can add data labels for that or
the audience eyeballs it. I change the number format from numbers to percentage
on the y-axis. Since this chart is about the adoption rate of technology, x, percentages are most
suitable to numbers. I also deleted the
background color and the chart outline. It's all about having color
and lines where they matter. The legend can be placed at the top so the chart
area is bigger. I also recall it the axis labels to make
them stand out less. Lastly, I recall it the data series to the
colors we always used. If you audience always uses the same colors
for certain items, be sure to do so as well. We want the audience to
focus on the message, and we did it by removing
any noise in the chart. The goal is to be a minimalist. Subtract everything
you don't need, keep everything you need. Don't add shadows or 3D effects. They distort charts and serve no purpose and they
just look tacky. Always remember
that less is more. You look at charts all day. Your audience doesn't
make it easy for them.
6. Refine Charts: Add Important Elements: Lesson five, adding elements. Now that we cleaned our charts, we have to direct the
audience's attention. We do that by clarifying
what our chart shows and add elements to
highlight key information. If we don't have a
visual hierarchy, the viewer doesn't
know where to look at adults around the chart. So we have to
establish a hierarchy to make sure our audience
can follow the story. First, we add a chart, titles. The audience should
not spend time wondering what
they're looking at. Show measurements if it's necessary in quantity
or in value. In this case, the y-axis
is in percentage, and it's clear from
that axis title. In scatter plots, you have
to add them manually. In other cases, the exits
doesn't need titles. The axes titles are
in bold to make them stand out against the
numbers right next to them. Data labels. Data labels are very useful, but can also lead to clutter. On the left, every
data point is labeled, but ask yourself, is
every point of relevant? Now, that is why we
should be strategic. You can also partly labeled
a chart as shown here. I have one label at the end and colored it with the
corresponding color. To make it stand out even more. It is bold, it looks
clean and precise. In other charts, you could
only labeled the highs and lows and the recent weeks
depends on your chart. Also consider how many decimal
points you want to show. You can combine data labels and the legend by displaying the series name in
the data label. Notice how it's colored as
well to its respective line. The audience can
not be confused. This is now the end product. Let's review all parts and their role in capturing
the audience's attention. We have a chart title describing
what we're looking at. The exits titles are in a lighter color since
they are not the focus. We have boiled data labels shown the last data point
and the series name. We substituted the legend
in terms of color. We have regency that stands out with blue because
this is the focus. Region a is lightly colored
since it's less important. By using colors, both text
and cleaning the chart, we established a
visual hierarchy. The audience can easily read the chart and it's
not overwhelmed.
7. Refine Charts: Highlight Key Information: Lesson six, highlight. We are still in
the design phase. Highlight key information. The audience knows
what the chart shows, and we want to
further refine how we show our data with
various techniques. Let's expand on the
visual hierarchy. I mentioned that region C stands out with
blue on the left, using a different color, direct the viewers attention. On the right chart, technology X stands out by being much
darker than technology. Why? Here I use to color contrast, use bold colors to direct the viewers attention
to key information. If we had more than
two categories, you could call it
the less important ones in a similarly light color. I'm defocus category
in a bold color. Animations. What you can do in
PowerPoint is to animate the chart
serious by series. As I show you here. Serious one, series
two, series three. The audience has less
information overload since the chart
appears bit by bit. It also makes the
story-telling more powerful. Text. Using text is also a viable
option to guide the audience. We give additional information
by using a textbox. Notice how I use colors and bold tags to make scanning
the text even easier. Another option is to recall
a data series one at a time. In our sales ranking, we could highlight the top two. Do that by reducing the
color contrast of data. We don't need to focus on Edit textbox to give
additional information. Then shift the focus to the bottom three products with
the same coloring scheme. Edit text to give context. This is a fancy way of
guiding the audience. We can do it by
adding animations such as fade in and fade out. And by grouping them, it needs more time to set up, but can pay off nicely. We can combine
multiple techniques together and create
a one-shot story. Let's take the sales chart from product G and walk through key
points in the sales cycle. The product G was
launched in February. Afterwards, we had the
large launch campaign on social media. In June, we had our
influencer campaign where people who are
showcasing our products. We had pretty good
baseline sales until this long before
Black Friday came up, consumers are waiting for
deals on Black Friday. Sales during Black Friday
went through the roof. Our marketing spend
was successful. During Christmas. We kept on pushing through all channels. And so with a lot. You see that by guiding the audience one by one
through each key point, you can add a lot of information without overloading them. We are done with
the design part. Let's look at things
to avoid next and then wrap things up
in the final lesson.
8. Things to Avoid: Lesson seven, things to avoid. The point of telling a
story with data and charts is telling accurate
information in an easy way. Ideally, the viewer only has to scan R charts and knows within seconds what the information
is about the message. However, that opens
up the possibility of unintentional or intentional
misinterpretation. We can manipulate charts to
make them look a certain way. Or we only use data that
shows us a picture we like. In any case, a clever
audience might catch a mistake and our whole story
and credibility is gone. There are certain
things we must avoid to keep our pride as
data analysts intact. Mistake number one,
incorrect, x is scarce. When we use the chart
about technology adoption, this case stops at 60%. Why not 100%? The goal of the chart is to benchmark the different region. It is only indirectly about showing the
total adoption rate, as we see on the right chart. In this case, our scale
ending at 60% is misleading. There is no reason to
stop at 60 per cent. Better child would
be this one with a scale to 100 per cent. We see how the
adoption rate rose, but we are still
far away from 100%. Choose to scale accordingly
to the message. This example demonstrates
an axis that is too short. On the right chart, the top product status
bar is cut off. The gap to the second product appears less they didn't is. Here is an example
with malicious intent. The poll results and
narrow with 51% yes, in 49 per cent, no. However, on this chart, it looks dramatically different. The reasons or the y-axis that doesn't start
at zero per cent. And the data intervals
are in 0.5 increments. This is a prime example
of a bad chart. Effects have to be
avoided to they distort shots and
serve no purpose, and they look dated. More difficult to be aware
of and rectify our biases. There are a lot, but the most common are confirmation bias, availability bias,
and selection bias. Confirmation bias
exists when you only take data into account
that supports your story. That might make the
story more cohesive. But what if you're falsely
mixing causes to effects? What if the audience has strong evidence to
disprove your story? After all, we are not here
to tell a simple story, but we present data
through a story. If our narrative gets
more exciting by introducing conflicting
data and address them, we not only have a
better narrative, but also the correct one. Don't shy away
from difficulties, just like everything
in real life. Data is rarely black and white. And availability bias arises
when you only take data into account that you think of right now or that is available. We miss crucial data. That could be the
correct solution. Instead of the data you
are thinking of, e.g. accompanies the smart sales
when temperatures alone, they conclude that
lower temperatures have a positive effect on sales. But they completely forget about the low
temperatures around big events such as Black
Friday and Christmas. If you take out
these two events, the relationship
could be different. This bias is especially dangerous since we don't
know what, we don't know. If there is something
unexplainable, it pays off to ask others in your team if they know
what might have happened. Again, don't mix false
courses with it affects, it can derail your story. A selection bias appears
if our sample data is not representative of
the whole population. Let's say you run a survey
with your customers, usually 30% of your sales
or premium products. However, in the survey that
numbers only ten per cent. That means premium customers are under-represented
and their needs are not reflected in the survey. Another pitfall is
missing context. Let's go back to the example we used in less than three
about chart types. The 100% stacked bar shows
the growing region C. But why exactly is every other region is
shrinking in size, meaning the market
is collapsing. Core is regency the only
region that is growing? By displaying another
chart such as the right one, we give context. In fact, the total
market is growing. That region seek
outperforms market growth. Addressing the
audience incorrectly can also be a huge mistake. We touched upon this example
briefly in chapter one, and it has to be repeated. Displaying one big concise
number might be wonderful for a Western audience
when they have to be introduced to a new topic. And too many numbers
are confusing. Showing plus 40% revenue growth over the last year
is a clear message. But what if the audience
is used to the details? They want to know the
historical growth in addition to a
breakdown by region, we would show a
heat map like this. The formatting helps
us spot outliers and we can further
highlight them with boxes. It gives much more context and satisfies our audience
who loves details. Misleading charts and biases, undermine your story
and credibility. Be aware of them and
try to avoid them.
9. Putting everything together: Lesson eight. Putting it together, we covered a lot until now and it's
time to put it together. After we identified the
audience and crafted the story, we can work on the visuals, which was good visuals with
the least amount of noise, and clearly identify
what they show. We redirect, abuse attention to key points and avoid
any tricks or biases. Now, we are ready to present at the beginning of
the presentation, describe shortly
what you will cover. It sets up the presentation and your audience has
an orientation. You either stay classy and
start with the exposition, or you could even start
with the solution, e.g. we should focus on
Regency technology x, and I will show you why. This is basically
starting with the hook. Then after the introduction to the three-part
story structure, set the stage, dive
into the problem, gives us a solution and finish
with a task or must know. Leverage your
presentation software and guide the audience
through the charts. Either use color
contrast, bold text, or boxes to highlight
key information. Don't present too much
information at once, or else the audience
stares at the slide. Instead of listening to you narrate your story bit by bit, you step-by-step animations to tell a story within one chart. Always think of
your audience and make it as easy as
possible for them. A clearly defined story
with suitable visuals and no unnecessary clutter makes
it easy to get the idea. After your presentation,
it pays off to send a follow-up email to
shortly summarize the presentation and sent
the PowerPoint file. Since you can narrate
over the slides, you should put the narrative in textboxes onto the slides. And the key information
is in the title. Because the audience controls the speed in which they
consume the story, you can pit more information
into each slide. When presenting,
it doesn't matter where slight begins or ends, but when reading the
slide on their own, each of them has
to have a message. Having 50 slides is not
useful for the viewer. Try cutting down the length
and your file to fit more visuals and information
into each slide. As you see in this example, we cut down for
visuals into one. With a small deck.
Your stake holders can repeatedly come back to the
story and think about it. After sending the e-mail, your story is complete. Any weight for the
manager's decision. Meanwhile, you can
craft your next story.
10. Conclusion: Well done. You've come to
the end of the course. You know the building
blocks of telling a story with data and are equipped if the do's and don'ts as each organization
is different, tried to adapt the
lessons to your needs. If you do that, you are more than ready to craft
your own story. Have fun, and always
try to improve. My name is Paul, and I
wish you all the best. Please rate the course
and visit my profile. Thank you for
taking this course.
Paul Madan, Data Analyst
