Welcome to Customer Analytics, I'm Professor Raghu Iyengar. I'm a professor of marketing in the Wharton marketing department, I've been here for about ten years or so. And during this time I've taught marketing research and other customer analytics related courses. So, today we talk about descriptive analytics, I hope your as excited as I am to talk about descriptive analytics. So, these days a lot of collection and analysis of big data is outsourced to third party companies who specialized these things. And throughout this module I'll introduce you to companies that do this kind of work in different industry segments. Such as scanner data at grocery stores, or metric for measuring audience engagement by media companies. But even if you're not involved in these industries, or you want to engage in descriptive data analytics on a smaller scale, it's very important to know about these companies. The way in which they're collecting this data, and the kinds of the questions that the data is trying to explore. Why is this? This is because it's a way of thinking about these kinds of issues. Thus, by learning about these companies, you will take away actionable techniques in thinking and forming questions, about descriptive data, techniques that you can employ in whatever descriptive data environment you are attempting to analyze. So what is descriptive analytics? Descriptive analytics can be defined in a variety of ways. So one is that descriptive analytics is a way of linking the market to the firm through decisions. Another way of thinking about descriptive analytics, it's the information that's needed to make actionable decisions. And yet another way is, it's principle for systematically collecting and interpreting data. What's the common thread here? The common thread is getting good data. But what I also want to talk about is that it's the synergy between data and decisions that managers have to make that makes for good analytics. So what are the different kinds of decisions that managers might have to make? One set of decisions might be purely exploratory in nature. So think about a brand manager, they're looking at their brand sales and suddenly they start dropping. Question is why are they dropping? Is it because customer preferences have changed? Is it because customers like competitors? There could be a variety of things going on. So in that sense, at this stage is purely exploratory in nature, we're trying to understand why things are not working out the way we expect them to. Another set of questions can be purely descriptive, for example again going back to the brand manager. I want to know what's my customer share of wallet? How much are they spending with me? How much are they spending with my competitors? Who are our customers? What's our segmentation like? So these kinds of questions require hard data in terms of understanding how much customers are, for example, buying our products or other competitor's products. Yet another set of questions can be purely causal. And the idea here is for example, if I'm changing the landing page on my website, how will it change consumer behavior? Would a change, would it in increase it in terms of click through rate, would it bring it down, and so on. So these questions, the one on the right most extreme, the causal questions, require systematic data collection and careful thought in terms of how to collect data. So what we see here is going from left to the right, the type of data that needs to be collected, the type of conditions that the data needs to be collected under also keep changing. So as we go on in this module today, we'll talk about different kinds of questions that manager need to answer, and what type of data is best suited to answer those questions. So let's begin with the exploratory type of data collection. So exploratory type of data collection is typically done to develop initial hunches or insights. Again recall the example that we started out with, the brand manager thinking about why the sales are dropping. It could be a variety of different reasons. And usually this type of data collection is a first step and very important step, to get a broad understanding of what the underlying problems could be. And it provides broad guidelines on what you should look for more rigorously. What's a typical technique that comes to mind when you start thinking about exploratory data collection? It's focus groups. Focus groups have been there for a long time. What's a focus group? Basically, you have about 8 to 10 customers in a room, usually you have moderator who designs the overall flow of the focus group. And you want these people to come and talk about the brand, their sentiments about the brand. Perhaps feed off each other, in other words you can observe dynamics, it's reasonably unstructured, it's free flow conversation. And what are you trying to do as a brand manager? Get insights into what might be some pinpoints for consumers. Now, in this day and age of data analytics and big data, focus groups have morphed, in some sense, to many different ways. Market research online communities, or internet communities, are basically focus groups on steroids. You can think about variety of different companies offering this service, for instance, Focal Point. What Focal Point does is basically rather than looking at 10 to 20 people, you start thinking about 100 to 200, or sometimes 500 people in a group. And you're monitoring them not for one time, but over a period of six months to a year. What's the idea here? The idea here is to build relationship with your consumers. Over time these 100 to 200 people, start building relationships with each other. They become more and more comfortable talking about their real feelings and real insights. Focal Point of course is not the only company doing it, there are many, many other companies. For instance, C Space is one of them and many other competitors as well, okay? There are many advantages of internet communities, one is that it enhances engagement with customers. So these customers are together, talking to each other, talking to the brand for about six months to a year. So clearly this closed concentration in terms of talking to each other, communication with the brand, it really enhances their engagement. Second, shorter deadlines are possible. Typically with focus group there are logistical issues in terms of trying to get these people in the room, get a moderator, and so on. Because you are looking at these customers for about six months to a year, you can actually have much shorter deadline. There are aha moments that come out, the most famous example is Kraft's 100 calorie pack. What did they do? They basically had a community, they had worked with C Space. They had a community that started looking at what do people want in snacks? What was the insight? It's not that people wanted to stop eating snacks, what they really wanted was snacks with low calories. Nabisco's 100 calorie pack has been an amazing success. There are caveats as well, what's the big caveat? ROI can be very hard to determine. Why? Because as you start engaging with an internet community, early on it might be quite difficult to forecast, what kind of insights will come out. For Kraft, this was great, but it will not be the case for every possible example. So, thinking carefully about when you might want internet communities, thinking carefully about is it worth the investment of six months to a year? Very important to determine as you start going into the idea of collecting data from customers, over an extended period of time.
