Fundamentals and Essentials of Business Intelligence, BI Course #1 | Michael McDonald | Skillshare

Fundamentals and Essentials of Business Intelligence, BI Course #1

Michael McDonald, Business Intelligence and Finance

Fundamentals and Essentials of Business Intelligence, BI Course #1

Michael McDonald, Business Intelligence and Finance

Play Speed
  • 0.5x
  • 1x (Normal)
  • 1.25x
  • 1.5x
  • 2x
7 Lessons (55m)
    • 1. The Basics of Business Intelligence

    • 2. What is Business Intelligence?

    • 3. Using Business Intelligence

    • 4. Preparing for a BI Project

    • 5. Applying Data Analytics

    • 6. Perils and Pitfalls of Data Analytics

    • 7. New Trends in Business Intelligence

  • --
  • 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

Business intelligence; the term is bandied about so frequently today that it might seem like everyone except you already understands it. The truth is that like its partner “Big Data”, business intelligence is actually not well understood at all. Most people who talk about business intelligence have at best a general notion of what it means and almost no experience with actually using BI in a meaningful way. However, BI is an important new tool for modern business. Advances in computing power can now give businesses ways to analyze data that they never could before. With these advances, firms can make decisions about pricing, marketing, new products, and resource allocation more effectively than they have ever been able to in the past. Major corporations like Kroger are starting to use BI to help determine what products they should advertise to specific customers. General Electric is using BI to more efficiently run its industrial maintenance schedules. Citi is using BI to help proactively figure out the maximum interest rates various customers are willing to pay. BI is useful in all of these settings and a lot more.

Meet Your Teacher

Teacher Profile Image

Michael McDonald

Business Intelligence and Finance


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.

Your creative journey starts here.

  • Unlimited access to every class
  • Supportive online creative community
  • Learn offline with Skillshare’s app

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. The Basics of Business Intelligence: Hi. I'm Dr Michael McDonald. I'm University professor of finance and the president of consulting firm Morning Investments LLC. Today, I'd like to talk to you about the fundamentals of business intelligence In particular. We're gonna be talking about business intelligence from a high level. What is business intelligence and why is it so essential for you in corporate America? 2. What is Business Intelligence?: Business Intelligence module one. What is business intelligence? Let's start by talking about decision making in businesses, So there's two ways traditionally of making different decisions in businesses. Whenever you're faced with a tough conundrum, you've two choices. First, you can rely on intuition or guessing or gut feeling. Call it whatever you'd like. But essentially, you're relying on the native ability of an individual to figure out a situation in the absence of any external information or data. The second way to look at making a decision is through data or past experience now. Traditionally, many businesses make decisions based on a combination of these two things. But at the end of the day, it's generally gonna come down, either on the intuition side or the data and past experience side. Most businesses tend to lean one way or another. The problem historically has been for most businesses. They either don't have past experience with many of their toughest decisions or they don't have the data. And so they're forced to rely on intuition, guessing, gut feelings, things like that. For instance, on a recent engagement I was on, I was dealing with a local ah brewery, and they were interested in potentially expanding through a new deal with a new distributor for their product. They would have liked to figure out what the prophet impact was going to be on their business. But they lack the data to go through and analyse that question. So as a result, they ended up having to make the decision based primarily on the intuition where the gut feeling off the president better data could have helped to address this question could have helped them to decide on what way to go with the decision in a more substantive way. Now, both these methods have been around for a long as business has. Each one has its own adherence. Not everyone is going to believe in intuition or gut feeling. They're gonna tend to be a little bit more scientific. They're gonna value data and past experience where other people are going to say that they want to trust their gut, and they're not necessarily gonna be interested in the data. After all, there is an old saying that if you pound the data hard enough, it'll tell you whatever you want to hear. Business intelligence is an opportunity to move beyond either of these two decision making processes on their own and instead focus on combining the two and in particular, enhancing the data and past experience method by using business intelligence, were able to go through and look at data in a more meaningful way without at the same time forcing the data to tell us what we want, where instead able to look at the data dispassionately, objectively in a scientific fashion to make better decisions for business. So why is business intelligence so important? Well, when it comes to making decisions, we have in every case a balance. We have opinion and we have data. In order to make these decisions, we have to balance between the two. We use our judgment, but we also need to use data as a way to figure out how our judgment should be related to a particular set of circumstances. Almost no one has the judgment to be able to assess a question that's completely foreign to them. But if we don't have direct past experience with a new issue, data can help us with making this decision. In the absence of this kind of data, business decisions are also are often made by the hippo. That is the highest paid person's opinion. We want to avoid that highest paid person's opinion being the sole rationale for particular business decision. Instead, with business intelligence, we're gonna use data to help drive our decisions, and that's going to give us increased confidence in the rationale for these decisions. So who uses business intelligence? Well, all industries have a need for business intelligence in some form or another. Today, some industries have traditionally used business intelligence for decades, so at least in a rudimentary form. Insurance companies are one example of this other industries. Today, though, even if they could use business intelligence or they're just starting till kind of look at this, there are often doing this in a very limited capacity. For example, manufacturing businesses today are often some of the last firms to adopt good data based, decision making good use of business intelligence. Most industries, though, are a mix of somewhere in between the two. They're perhaps not quite as data centric. His insurance companies, which have been using the data for decades to help derive insurance premiums, but they're not necessarily quite as unfamiliar with business intelligence as many manufacturing firms are so industries like banking, for instance, technology. They have some experience with data, but perhaps not quite as much as they could. The goal in this presentation today is to help your firm and you personally understand how you can use business intelligence in your day to day job. What is business intelligence? Well, business intelligence is essentially just using data to enable us to make intelligent, fact based decisions and thus eliminate guesswork. To do this, we have four steps that we need to follow. First, we need to gather and clean our data. We'll talk more about how to do that later on, as well as each of these steps. But in general, all businesses have access to some data. It's just a matter of gathering that data together in the right place and then cleaning it to make sure that the data is actually accurately representing what we want it to. Next. We're gonna go through and analyze that data. They're very specific techniques that we need to use to analyze this data. In a future course, I'll walk you through those techniques so that you can see step by step. How do you go from a spreadsheet full of data to useful insights. Step three is to then go through and test our choices with data. If we're faced with a business decision of some sort, we need to go through and use that data to figure out what the impact on our company's profitability or sales or costs might be. Based on the data that we have available. Finally, using that data and using the insights we've derived from it, we can go through and make a decision off course weaken. Supplement this with our best judgment. We can understand as an example the cost, tradeoffs, foreign investment and the risks that we're taking on in a way that simply looking at the numbers won't let you dio. But having that data in front of us gives us a lot more confidence and a lot more certainty that we're making the right decision. Then, if we didn't have anything to help us, what is Business Intelligence, though now? Originally, business intelligence was coined by Gartner Group in 1993 but it's often simply described as a collection of pieces of software. It's not business intelligence and said is a set of technical techniques and processes involving statistics that are gonna let users analyze data and make decisions based on that data. This can be done through specific software programs, but it could also be simply done through excel or generic programs. In theory, business intelligence could even be done by hand on pencil and paper. Now, of course, that would be cumbersome and unwieldy, so software helps us to automate the process. But at the end of the day, the danger with relying too much on software is that you're creating a black box, which you may or may not understand. It's good instead to have an understanding of the actual processes and the tools that are involved with business intelligence so that we can go through and make Thebe best decisions possible. In a nutshell. Business intelligence simply means that we're taking raw data and we're transforming it into useful insights. Business intelligence decisions can be used for a variety of different types of questions, though, for instance, how should products next to cash register be priced in order to maximize our profits through impulse buys for shoppers that are checking out from our store? Another question we might look at what customers should we offer discounts or coupons to? We could offer coupons to all customers, but that might result in a lower average price or lower revenue overall for the company. Then, if we only offer those coupons to targeted consumers that are most likely to use those coupons but wouldn't use our service in the absence of coupons, 1/3 question might be what piece of equipment in a factory will fail next. We might have lots of different pieces of machinery out there. Various parts in these machines might fail from time to time. We can use business intelligence to go through and enable us to predict what parts and pieces in a machine are likely to fail, and thus to cut our maintenance costs and avoid unnecessary expenditures on maintenance. Another question you might face which borrowers for a bank that you work at are most likely to default. We have lots of different customers that come into a bank looking for a loan which ones are the most likely to default in our existing portfolio of loans? Are there any customers that we should be particularly concerned about and proactive in terms of reaching out to them or being prepared to modify their loans if necessary to deal with the potential for default. Another question might be given trends in the economy. What are our projected cash flows from a business? This is an issue all companies have to deal with. We have some set of past historical figures, and we need to figure out what future cash flows are gonna look like. But we don't know exactly what the future's going to look like. It's certainly not gonna look precisely similar to the past, so as a result, we need to use business intelligence to help us with this cash flow forecasting. Finally, we might ask, Where should we locate a new store or office location? Decisions are an easy opportunity for businesses to use business intelligence to help them optimize their investments in the future. 3. Using Business Intelligence: module two Using business intelligence Business Intelligence, as we saw in the last module, is useful in a wide array of different types of questions that a business might face and a wide array of different industries and circumstances in particular. We can kind of break down how an enterprise makes decisions across many different groups. For instance, marketing departments are concerned with pricing, promotion, loyalty, things like that. All of these air driven by the demand for the product, which is ultimately a function of our consumers or customers in our production capacity, we might be interested in our output capacity are labor costs or labour capacity. Materials available are materials costs. Again, all of these are gonna be driven by the quantity of output that we want to produce, and they're related to the suppliers who supply these materials for us. But these are all issues where we're gonna face decisions that big data can help us to solve. Third in the finance vertical, we might be concerned about cash flow or debt to equity or investments. Which type of investment should we take on as as a company? What type off financing should we use for a particular investment that we're looking at. How are revenues gonna change over time under certain circumstances? How will our profits be affected by a possible set of decisions that we face? All of these questions are central to the finance, vertical and really part and parcel of the CFO's job and challenge. These questions are critical to investors in firms and to equity owners, business, intelligence and big data. More broadly can help us with all of these issues. So how should we go about using business intelligence in business? Well, when we're looking at company, we often face to competing challenges, uncertainty, meaning, what can happen and complexity. How do we deal with all of the myriad array of variables that we're faced with and their impact on our business? We're gonna address each of these challenges in different ways. The uncertainty challenge we can address through having good data. We're gonna take a lot of different data that we have available to us and put this into our business intelligence function. That's gonna help us to reduce our uncertainty if we don't know what could happen. Potentially gathering and looking at data from other circumstances be historical data or data from what's going on in the world right now can help us to reduce this uncertainty on the other side of the coin. In terms of understanding the complex set of variables that we face, we can come up with a model that's going to represent our behavior and our need for a system. I understand all these variables. We put these two things together, the data and our models, and we're gonna be able to come up with an intelligence set off outputs that will let us go through and designed the process that we need to make choices. On the data side, we're gonna look at variables like measures and estimates. For instance, we might be able to measure how much our historical sales have been. But we could also make estimates about what our sales might be in the future, and we could do this as an example on a customer by customer level or a division by division level or even at the company level. We can do this for US competitors, and we can do this for our suppliers. Off course. There's gonna be some probability in some uncertainty in here, but the more data we gather, and the more estimates we make, the better off we should be in this face on the other side of the coin, models are gonna come down to structuring our relationships properly, representing our problems appropriately and then generating alternatives. For instance, if we're trying to make a decision about where we should locate a new facility, we might face a question about where our customers going to come from. It's not possible to represent all of our customers and their individual locations at any given point, but through big data and through appropriate models, we can represent where they are approximately and so their travel time to our location, for instance, and the relative convenience for clusters of customers that were interested in. We can do all of this through spreadsheets or through software. But either way, putting together these data and the models well, let us come up with more intelligent choices for the decisions that we face. All right, so perhaps you're sold on the value of B. I. Certainly we all face very complex decisions and choices every day. So what do we do now? Well, there are certain steps that we need to contemplate for completing a B I project. In particular, Step number one is to gather data. Step number two is to clean up that data data itself is not necessarily always going to be perfectly clean. And so we need to go through and make sure that the data is accurate. And that represents the question that were interested in Step number three is gonna be to run an analysis on that data and step number four is to test our choices or options based on that analysis. For instance, this could be a simple as gathering our cost inputs and are expected cash flows and running a NPV calculation on a couple of different choices. Finally, fifth, we're going to make a decision based on the analysis and based on the outcome from our different tests and choices. 4. Preparing for a BI Project: module number three Steps in Business Intelligence. All right, so let's go through and look at each of these five steps in detail and talk about how you'll go about accomplishing them, at least from a high level step. Number one in using business intelligence is to gather data now. Lots of companies say, Well, we don't have access to data, and I'm not sure where we could get it, So I don't think business intelligence is for us. That's a bad decision. All businesses have access to lots of data. Whether you recognize it or not, your business has access to lots of data, whether you're aware of it or not. Now, sometimes that data's proprietary things, like our customer data. Sometimes it's publicly available data, for instance, data on household spending patterns from the U. S. Census Bureau or on economic activity from the Fed. The government generates reams of data. It's gonna be very unusual that we want publicly available data or proprietary data alone in isolation. Instead, were often gonna wanna pull pieces from both of these things to help inform our decisions going forward. For instance, if we're trying to if we're trying to forecast our cash flows in the future. We're certainly going to be interested in what our historical cash flows are and the number of customers that were generating and the number of sales people we have in our projections for the future. But we also want to take into account forecasts from the government about what the economy like might look like, or forecasts for our particular industry from the government. We might want to take into account historical data and what the business cycle looks like. All of these things can give us a more accurate view of what our profits will look like in the future. The UN result here is that the data that's needed depends on the question being asked. A mix of public and private data is gonna work very well forced in most cases. But at the end of the day, we need to gather that data so that whether we're talking about choosing a price for a new product that we're going to offer or choosing a site for a new location, we have access to the right kind of tools for the job. Step two and business intelligence is to clean up our data. To do that, we're going to need to go through and look for issues in the data. I like to say that data is a lot like data are a lot like kids. They often get dirty. Data can often contain outright errors. For instance, in a recent project I was involved with, I was looking at a set of investment decisions by a NASA manager, and in some of the data, I notice that they had transposed the ticker symbols for individual stocks and the gross margin of those stocks. As a result, if you are trying to run an analysis, half of your data would have been bad because you'd be looking at thes ticker symbols, which are alphabetic codes, and trying to run a numerical analysis on it. That transposition was buried about 1/4 of a way through the data set. But if you hadn't observed that if you hadn't gone through and cleaned up the data, 3/4 of your analysis would have been useless. It's also problem because data availability can change over time. For instance, S I C codes and N ai CS codes. Both track individual level industries s I C codes, for instance, are four digit codes that track companies based on their particular industry. In the S, I C code s I C codes were replaced starting in the late 19 nineties and early two thousands by any I CS codes. If we're trying to look at a couple of decades worth of data, we might find that we have a mix of S I C and N ai CS codes. If we don't distinguish between these two, we can run into significant problems over time. In particular, let's pretend we're interested in doing some sort of forecasting related to the current oil price slump. Well, you might say that the last time we had an issue this bad with oil prices is in the mid 19 eighties. So if we're trying to compare 15 16 6017 to the mid 19 eighties, we can do that. We need to make sure that our data availability is the same between those two periods. A Sigh C codes and any I CS codes are not easily equitable, so it would require a little bit of manual work to go through and clean up that portion of the data. Further data can often be marred by unusual or un observable effects. For instance, we could have a situation where we had look at the number of customers coming into our store every day. But we have one customer that comes in on a particular day, and they happen to spend 10 times the normal amount that the average customer would that can throw off our revenues for the day. And so if we're trying to forecast on a store by store basis what our sales should be for the future, that data point could throw off our calculations. Thus, the issue here to solve is how we condemn eel with these unusual effects. And the answer is that we want large sample sizes in our data going through and cleaning up the data is great, but we wouldn't want to throw out that customer that ordered an enormous amount of volume. We don't want to throw him out of our data set. He is useful and they do. You do have customers like that periodically, but at the same time, we also read need to recognize that these customers are infrequent. Ah, large sample set of data helps us to address this problem. We want to clean up our data through testing. At the end of the day, the first step to do this is gonna be to start by going through and looking for out liars and then wins arising those so with a given set of data, you'll want to go through and look at the normal distribution of that data and then throw out those observations that don't seem to fit the mold or at least be aware of them and set them aside. Create some sort of marker that will tell you when some of your analysis might be impacted by those unusual observations. We'll talk in a future course, but specifically how to clean up our data on a step by step basis. The issue of data leads to the problem of information governance, though in particular many people have heard the phrase garbage in garbage out or guy go Now , information governance is supposed to help us to prevent these types of issues, in particular if we have bad data going in. If all of our data is an example, comes from employees manually entering transactions at various facilities, there's a good chance of those employees might simply write something down wrong. If employees air recording these things by hand, our data going in is gonna be garbage. Which means that our data going out will also be garbage. No amount of analysis or data cleaning can fix bad data. All we can do is throw out bad observations instead. If we go through and have a very specific set of processes in place, we can help to ensure that our data is gonna come out cleaner and our analysis and decisions. We better in particular we should focus on the quality of our data. We need to go through and hold accountable people who are responsible for collecting these data for their accuracy. This could mean customers could be in vendors. It could mean managers within the firm. There are a variety of places that are data can come from, but we need to do everything we can to make sure that the data we're getting are accurate. If we get accurate data than through business intelligence, we can create significant business value 5. Applying Data Analytics: our third step in business intelligence is gonna be to go through and run our analysis. Once we have the clean data set, we want to go through and run that analysis to figure out what those data mean for us. There's several different methods that we can use to do this. A very simple method would be to simply look it means or medians and look at differences between them. For instance, if we're trying to compare two different groups of customers and figure out who we should target our advertising efforts to, we could look at simple mean or average differences in spending patterns between the two groups. Or we could look at median differences in spending between the two groups. Then we could figure out what those differences are and do some sort of a statistical test like a T test, determine those differences and how they relate to the two groups. That type of analysis is useful and easy to do, But the problem is that it doesn't control for any other factors might be influencing our data a way to avoid this issue and to take into account other a myriad array of other factors that could influence our data is through regression analysis. Now there's a variety of different modifications we can use over simple regression analysis to help us deal with problems that might come up. But all regression analysis is simply based on the idea that we're going to try to account for other factors that might drive our decisions. For instance, if we're looking at a pricing decision, we might not just look at demand for our product. We might also look at what's going on in the overall economy. What day of the week is it? How much advertising have we done? What kind of customer service are we running? There's half a dozen variables that we could look at very easily, all of which might plausibly drive our pricing decision. We can take these into account with regression analysis in order to figure out what are optimized prices, little, maximize our profits. Now, perhaps we don't have good data related to the issue that we're trying to look at, or even if we do have data. Perhaps we only have a small amount of to get around this issue. We might look at Monte Carlo simulations to look at a Monte Carlo simulation. We can go through and weaken form on idea of what our data should look like. Based on other data we collect from other sources. Ah, Monte Carlo Simulation is simply a way to go through and hypothetically create data points . That represents what the outcome of our testing could look like. For instance, in our pricing decision, we might go through and create hypothetical demand data points based on pricing patterns that we see now. You might say, Well, all of this hypothetical sound. Okay, but where's the rial data? Monte Carlo simulations are based on the models that we see in the real world. This is where the models come into play. So if we can accurately model the behavior of our customers are our suppliers, then we can use Monte Carlo simulations to come up with a range of possible outcomes from what might happen in the future and to make sure that we're prepared for those outcomes. Finally, some more advanced techniques that we might use an analyzing our data are network analysis and control functions. Each of these again are different ways to take into account. Other factors might be influencing our decision making process. Let's take a look an example of this. For instance, Let's pretend that we're simply looking at college in admissions criteria were interested in figuring out which people we should admit to a hypothetical college. We have a variety of data available. All our applicants have to put in S a T scores a CT scores AP exam scores, etcetera. They're also gonna have to report their grade point average their class rank, and we'll have a metric for the strength of their high school will also have more subjective data. Things like whether they're in various extracurricular activities like band choir clubs, sports, any non school activities they might do. Like working, volunteering community groups. Etcetera will also have an idea of what they'd like to do at the particular college their intended major and finally will have an idea of what their family legacy is like. Have they been to this college before? Do we have any experience suggesting that they might do well based on either their family or their home, state or country to figure out the likelihood of success in college through a metric like, say, predicted college GDP, We could go through and create a regression analysis. That regression analysis essentially says, What's the likelihood of success in college measured by saying GDP? And we set that equal to high school graduation that is a one or zero variable, whether they've graduated high school or not. Time, some sort of waiting factor for that plus, for instance, their GP a times a weighting factor for G P A. Plus a variety of other factors multiplied by weighting factors. We can go through calculate thes weighting factors and then, given these inputs, figure out the likelihood of success for any college student entering any university in the country. The same type of analysis can be done in your business. We can go through and figure out as an example what your revenues might be forecasted to be based on a variety of internal and external metrics. We could go through and figure out what your profits should be. We could go through and figure out what your optimized price might be. What would your profits be under given price differences? What would your profits be if we make various investment decisions again, all controlling for a variety of other factors? That's the difference between business intelligence and traditional ways of thinking about things a traditional business decision might mean simply saying, If we do Project X Y Z, it'll add $5 million to our bottom line. Business Intelligence lets us look at how Project X Y Z will interact with. All of the other factor is going on in the company and all the factors outside the company to help us figure out what our profits will actually be next. When making a decision involving business intelligence, we want to test our options. Business intelligence today is often used just for monitoring and identifying problems, and that's great. Or at least it's a start bought Business Intelligence is also great from making decisions. I've talked multiple times about as an example how business intelligence can be used for determining pricing. But business intelligence can be used for a myriad array of different decisions out there. A good business analysis model should let us test ah, variable of interest based on various choices, any variable that we like. For instance, how will our profitability be impacted if we give various employees a raise? If we give different employees a raise? There's of course, immediate cost in the form of higher compensation. But we also hope that those employees will work harder. Based on that, that's the point of giving them a raise. After all, a good business intelligence model should let us go through and figure out the profit impact from giving the employer race. Figure out how their productivity will be impacted by that raise. Another example. House profit impacted If we double our marketing spend again. There's an upfront cost, but we need to understand how's that cost gonna pay off in terms of increased demand, increased sales, etcetera. Business intelligence is useful for this type of issue We want to go through. Take the analysis that we ran in our previous step and test our choices. If we're doubling our marketing spend, we could go through and use our regression analysis and predict profits with, say, marketing, spend of $1000 per customer per year and marketing spend of $2000 per customer per year. Finally, based on the results of that testing decision, makers like you have a basis on which to make choices. Going back to our marketing example for a minute, we can figure out what our profit is if we double marketing spend versus if we leave marketing, spend static again after taking into account all other factors that are out there like the state of the economy. What else is going on in our company, etcetera. Then we have a basis for making decisions. For instance, is this investment in marketing dollars worthwhile based? Unexpected profit is our pricing set optimally so that we're gonna maximize revenue are our resource is allocated efficiently? These types of decisions are going to be best answered with data. Now, of course, with any of these things, our testing is never going to be 100% perfect. There will always be some risk involved. Over and above that, we could have something that's unexpected that we didn't include in our model that goes wrong. We might, for instance, double our marketing spend expecting that it's going to help our grocery store. But all of a sudden, a new competitors that we never expected moves in across the street. That wasn't in our model, and it couldn't be predicted. That doesn't mean our initial decision wasn't the right one, given that the information we had at the time business intelligence helps us with making these decisions and controlling our uncertainty, 6. Perils and Pitfalls of Data Analytics: module, four advantages and Disadvantages of business Intelligence. So we've talked a lot about business intelligence and how it works from a high level overview. In a future, courses will go through and drill down, step by step on how we implement business intelligence in a variety of industries. But for now, let's talk about the advantages and disadvantages of business intelligence. To begin with, Business intelligence has significant advantages these days, even with modern computing technology and the Internet and everything else. Many decisions made by businesses are largely guesswork. Business intelligence is transformational, and that at least gives us some basis in past experience for making decisions. If we're faced with a choice we haven't confronted before, we can rely on data and statistical analysis and testing for making those decisions. The statistical testing and analysis doesn't have to be impossible. It doesn't require we hire somebody special to do it. We can do it on our own, and this gives us a tool to address these issues further with some of our advanced business business. Intelligence techniques forecasting and decision making have been shown to be much more accurate and much more effective by researchers looking at these issues, There have been several studies done by researchers looking at management decisions, accounting decisions and financial decisions that have shown the company's air more effective and make better decisions when they rely strongly on data rather than subjective guesswork. Third, Business Intelligence helps with a key age old problem in business continuity. In particular, we might have a great CFO or great CEO or a great chief marketing guy or a great vice president, but if some of these people leave, all of a sudden we're faced with new issues for the company. We've lost wealth of experience, and that creates problems for us. How are we going to deal with that key man? Risk Business Intelligence helps to make sure that we're making decisions objectively based on data and based on a very clear cut set of processes and procedures. Rather than making decisions subjectively based on hippo, the highest paid person's opinion. This leads to consistency in decisions Over time. I'm gonna say that last sentence again. Thus, business intelligence leads to consistency in decisions over time, business intelligence, disadvantages. Now I'm not going to sugarcoat it for you. Business intelligence also has significant disadvantages. First of all, it's easy to look to lose sight of the big picture when we're talking about business intelligence. Instead, many people become so focused on particular data points or particular sets of data, they lose sight of the big picture. That could be a real problem. It's important to be able to holistically, take in this set of data, extract conclusions from it, and then establish how those conclusions impact our overall business. There's also some costs associated with business intelligence. We certainly have a cost in terms of time, potentially in terms of personnel hours. Certainly in terms of some data, whether it's internal or external data, we have to gather that dated a minimum and potentially some software. As I mentioned at the beginning, business intelligence can be done in excel in many way in many cases. But if you want to make it a little bit easier and faster, you may end up investing in a software package. There's a cost associated with that. We can also confront the classic problem of analysis paralysis. I've seen plenty of clients out there where they're faced with reams of data, and they're not sure where to go with that they go through in the analyze the data, and they look at the study, the options so much and can never reach a cogent conclusion. So they never end up making a choice. They sit for prolonged periods of time and just end up going with status quo rather than making a proactive decision that will help their business based on the best available data at the time. On the other side of the coin, it's way too easy with some software packages to become wrapped up in the technology to become reliant on the black box. A lot of software packages out there make basic business intelligence very, very easy to do. The problem then becomes, at least for this basic business. Intelligence is being done. You don't have any idea what's driving those decisions. You're just seeing the output rather than seeing the process along the way. Sometimes it's important to see how the sausage is made. In particular, we want to know that the data is being crunched the right way, and then the output we're getting is valid. This is less of an issue with advanced business intelligence, since most software packages out there aren't good at advanced business intelligence type functions, at least not without significant input from a human. Let's talk about business intelligence technologies for a minute, so the business intelligence industry is consolidating over time. There's a number of different analytic databases out there, for instance, starting with, say, the grandfather in the space Oracle. But there's lots of competitors thes days. Business intelligence is definitely a consolidating industry, though. Oracle S A P I B M E M C, Hewlett Packard, Terry Tate, All of these companies are consolidating and looking to take on and gobble up smaller players in the space and then to sell you holistic packages of software than theory are supposed to go from start to finish. There is still some independent vendors out there, like MicroStrategy in from Attica, SAS, etcetera. But ah, lot of these different vendors, as I mentioned, are consolidating their offerings. The problem becomes that because they're trying to make the technology so easy, it becomes easy for users. You and the people working with you to end up not really understanding the business intelligence. It's being delivered in particular. It's very important that you understand exactly what you're getting and you're using the right tool for the job. A black box technology system might seem easy to use, but you might be using the wrong tool and thus getting the wrong output or even making the wrong decision. I guess at the end of the day, the end result is, technology can often be overrated. It's more important to understand how to use business intelligence techniques than to have the right software. Almost any software package can be used to deal with any different business intelligence technique. I've used a variety of these packages, and you can use any of them for most issues. There's not a good reason to use one over the other. It's generally pure preference. Instead, the rial skill and the real advantage is an understanding. The underlying business intelligence technique said. You understand when you're using the right tool versus the wrong tool 7. New Trends in Business Intelligence: module. Five. Trends in business intelligence Business intelligence is changing rapidly. The term was hardly ever used a decade ago. Today, it's bandied about all the time. So where is business intelligence going in the future? Well, there's a few key trends that I want to spend a couple of minutes talking about. First is the cloud, so everybody's familiar with the cloud at this point. But the cloud is changing business intelligence because it's improving the speed of decision making. If data is input and uploaded to the cloud, then we can get that data in real time rather than having to wait for a weekly ah monthly or a quarterly report. As a result, if we analyze our data in a real time way, for instance, we adjust Our pricing for our company is an example. On a daily basis. In real time, we can maximize profit while at the same time taking invention of the day that we have available to us. This might sound absurd who sets their pricing on a daily basis, But the reality is that demand for products changes over time. By waiting a week or a month or 1/4 we might be delaying responding to that demand, particularly if we're talking about an Internet based business or technology based business where many of our transactions air done virtually. There's no reason not to use cloud based decision making to change our prices frequently. Many of the major airlines and travel sites, for instance, alter their pricing on a day to day basis. We can use data to deal with these types of issues. Next is social media. So there's new types of data, like textual analysis that are giving us new opportunities to figure out interesting questions. For instance, we might be looking at our businesses Facebook page on. We have all sorts of comments on here. It's easy to identify the comment that's really negative, and maybe we can take steps to address it for particular customer. But some comments are going to be just a little negative, and perhaps we overlook them. Some comments might be kind of neutral, and perhaps we overlook those certainly some comments that are only somewhat enthusiastic rather than truly exuberant. We might not bother respond to those. We can use textual analysis with a dictionary of words to put a numerical score on a particular comment by a customer. That numerical score gives us a way to quantitatively assess that customers experience and then to take the appropriate action, for instance, to send them advertising in the future to try and get them to buy our product again, or to send them a coupon to try and ameliorate a less than stellar customer experience. 1/3 major trend in business intelligence is Advanced Analytics. In the last five years there, several new waste analyzed data that have been developed. And thanks to improvements in computational power on everyday computers, it's a lot more feasible to use traditional ways off analyzing data that hadn't been feasible in the past. In particular, advanced analytics rely on things like regression analysis and Monte Carlo simulations and control functions. Those types of technologies those types of mathematical functions hadn't necessarily been feasible to run on computers a decade or two decades ago. Today, thanks to improvements in computing power, that's much more feasible. So what's business intelligence today versus tomorrow gonna look like? Well, one often discussed quote is that business intelligence today is like reading a newspaper that his business intelligence is just a reporting tool it's on top of a data warehouse. Maybe it loads nightly and it produces historical reporting. Business intelligence Tomorrow should focus on real time events and predicting tomorrow's headlines going forward and helping our customer moving a proactive direction to deal with challenges before they emerge. Rather than be reactive fashion in the advanced Analytics space, there's a few different issues that we might be interested in. In particular, Advanced analytics is often equated to predictive analytics. So if we're trying to predict the future, we need to go through it. We need to have a robust set of functions that are going to let us mind our data so that we understand exactly what we're doing and what kind of data we have access to. Then we're going to use powerful regression toe tools. They're simple regressions that we can use. But if we modify those regressions a little bit, we can go through and better understand the choices in the options that we have. Third is Monte Carlo simulation. As I mentioned earlier Monte Carlo simulations have improved dramatically in the last few years in terms of their predictive power, and fourth, we need to be able to go through and look at which data are statistically significant. In our particular context, this means testing our data as it comes in to figure out which data points we should be focused on. Then, finally, we can combine some of these tools to go through and predict our customer behavior. We can look at things like churn and a Torshin. We can look at purchases we can go through in profile our customers to figure out which ones are most interested in our product and which ones are least likely to come back in the future and how we can address those issues. In a related capacity, Retail Analytics is a major emerging trend in business intelligence market Basket analytics . Let us do all sorts of powerful things to help our customer weaken go through, for instance, and assess which products a customers buying and potentially offer them add on products they might be interested in. This can be done using either traditional data analysis or text analytics. Justus. We could use textual analysis to gauge customer enthusiasm or customer experiences based on their comments. We can also go through and use text analysis toe. Look at our customers behavior in a retail setting. Further, we can go through and cluster our customers and segment them into different groups that all let us set pricing and all and advertising offers in a way that's appropriate to their particular customers. Some of our customers are gonna be more price sensitive. Others arm or focused on customer service. We could go through in segment our customers so that the customers that are more focused on customer service we devote more resource is from customer service to them and the customers that care more about pricing. We devote more coupon ing and advertising efforts to them. This leads the idea of tailored product assortments, so we could certainly go through and look it for a different group of customers. What types of products are they gonna want to combine and not a lettuce maximize our sales ? This also helps us with the issue of inventory forecasting. If we can better predict what our customers are gonna buy, an increase our sales. This helps US forecaster inventory for the future. Inventory forecasting is a very useful business intelligence tool. Whether we're talking about retail or any other space, the let's look at a couple examples where business intelligence is useful. In particular, let's start by looking at amazon dot com and Netflix. Amazon dot com and Netflix have opportunities where they implement collaborative filtering and try to pick predict other items that a customer one might want to purchase based on what's in the customer shopping cart and the purchase behaviour for other customers. This is a classic use off big data. Amazon looks in real time what's in your shopping cart and then goes through an adds other products suggestions based on what they think you might want to buy. This increases the opportunity for impulse buys. Netflix does exactly the same thing with shows that you watch public text analytics. What is Text Analytics? We'll text analytics simply taking unstructured customer comments and turning them into actionable insights. Another words were trying to go through and find nuggets of insight in our text data that's going to improve our business. Essentially, Wikipedia is going to describe this as linguistic, statistical machine learning techniques that model and structure our information in ways. Let us do a better job with our business, and that's great. But at the end of the day, text analytics really simply means taking something that's inherently on opinion of a customer and thus is very difficult. Analyze in a data centric way and turning it into some sort of a qualitative metric that we can go through and evaluate in a numerical setting. Let's dig a little deeper on this, so let's envision that we start with a common that's uploaded through Facebook page, Twitter page and email, etcetera. The customer goes through and puts in this common. The traditional way of dealing with this is to read the comment and send him a thank you note or an apology note or something like that. Instead, with text processing weaken. Do that. Thank you note or the apology note, of course, but we can also take that comment. We can go through and categorize it based on a variety of possible different categories. Is this service related Is a quality related, Is it cost related or friendliness related? What is the real issue that the customer is addressing? Then we go through and tune that comment. What metrics are we going to use to evaluate that customers experience? How did the customer's experience stack up compared to other customers who've talked about , say, our quality or cost or friendliness. And then finally, having assign numerical scores based on the objective words of the customer used, we can go through and analyze this data by the thousands or tens of thousands to create alerts and real time actions. For instance, if we're looking at Starbucks, we might look at their Facebook page, and we can go through and isolate specific quotes that help us to figure out how our customers air feeling. We can use this not just for Starbucks but also for places like McDonald's. For instance, we could look at their new Metcalf a shake, and there go and talk about how our customers are responding to that McCafe shake weaken, go through fine text that matches that and then find the customers that are responding to that and the way they're responding. For instance, in the case of Dunkin Donuts, we have this customer who announces that they have discovered the new Dunkin Donuts Starbucks coffee shortcake. Yum didn't buy the toasted almond flamer that was there, too, but should have. We can go through and establish different product categories in this case, the strawberry shortcake and the toasted almond flavor coffees, and then the comments that are associated with it. In this particular case, this individual apparently likely product. We could assign that a different score than someone else who didn't like the product. And we can use that to assess. For instance, local marketing, perhaps strawberry shortcake Coffee does very well in one geography, but not so well in another. By going through and analyzing this text in a numerical way, it gives us a tool toe. Look at the data in a more meaningful sense than if we're simply read the comment and respond. This concludes our overview discussion of business intelligence. I hope you've enjoyed this presentation. Thank you for watching. I'll be up loading future hands on courses in business intelligence techniques coming soon .