[MUSIC] Quick recap of what we've been doing so far. So we did factorization of data, we did clusterization of data. It is now time to do the third piece, visualization of data. And the data we're talking about are perceptual data. Just makes it a little more interesting. So let me, in some sense, walk you through an example just to get you used to what it is we will be doing right away. Imagine you're a stationery retailer called OfficeStar. This is actually an example coming from ME Excel, okay? Designed by folks over at Penn State. You're a stationary retailer, say the name of this retailer is Officestar. You want to know how you are perceived in your town, vis-a-vis, three main competitors on five key dimensions. Just imagine the amount of work that has already been done. One, you identified who your competitors are, two, you know what are the dimensions on which you are compared with them, okay? That is a lot of work, but assume this has been done already. So well those are those five dimensions, things like large choice, low prices, service quality, product quality, and convenience. There is a sixth dimension called overall preference. It is meant to catch dimensions that we might have missed. Okay it will catch them. And the three main computers apart from OfficeStar are Paper &Co, Office Equipment and Supermarket there. So I have six attribute dimensions and I have four brands. Totally what I will have are actually 4 x 6, 24 data points is all I need to build my perceptual map, 24 data points from every single respondent. How do I do this, okay? How would you collect this data? Say you survey a small number of people, it doesn't have to be large, even ten respondents is good enough to start with. In the target population, and how they rate each firm on each dimension. So you have to ask each respondent 24 questions. And then you average the scores across these respondents, these are Likert scales by the way. And we project this in our object, this five dimensional object that we get into perceptual space, two dimensional perceptual space. That's basically what we will be doing now. What does it look like? So when I actually do the perceptual mapping, this is what it looks like. I have a two dimensional screen, on which I'm actually projecting five dimensions. All those five dimensions. And those dots that you see, Those are basically the branch. Where the branch are located vis-a-vis the dimensions. How do you read a perceptual map? I'll put out the instructions for how to read them out over at Coursera and I want you to do that. So basically that and there are some questions based on how to read a perceptual map. So I'm just going to put them up there. How is the perceptual map created? It is again coming from a factorization of data, basically. And the desktop app will do it for you. The Which I will supply. Somebody of insights, just by looking at their desktop that desktop generated map. What can we say? Well a lot of things. How does it help? You've heard of a SWOT analysis? Strength, weaknesses, opportunity, threats. A perceptual map like that can give you an understanding of strengths and weaknesses of your brand vis-a-vis others. How are they placed in each of those dimensions? It gives you the O in opportunity. Basically, tells you are there white spaces where entry could happen? Other places where the other guy is stronger than I am. If they try to muscle into my space, how will they move? In perceptual place, how will they try to move? Then, there is the T part, the threat, in terms of potential entry, Where, in some sense, can other competitors kind of try to enter my domain? And in this case, you can see each respondent is basically rated each of the specific attribute. The question then is, how do we bring overall preference into play? Well, I'm doing that. When you bring overall preference into play, you get what we call joint space map. So we have the original perceptual map on that. Those magenta lines are basically individual preferences of every single respondent, right. So ten respondents, you can see what's going on. Okay, the note given will tell you how to read a perceptual map, how to read a joint space map, and I have a set of questions for you that I would like you in some sense to attempt. The answers for which will be provided separately so you can compare and see. Mapping consumer perceptions off brands. So, basically, putting it all together, good positioning requires a managerial understanding of the perceptual dimensions which best describe the market. And that are meaningful to consumers, both are important. Positioning of existing brands along these dimensions. The relationship between physical characteristic and perception. Airline IndiGo might say we offer a lot of leg space, but if passengers don't perceive it to be true, then you know there is a problem, right? So all of that can be seen, and consumer preferences. Even with a small number of respondents. And this is the bottom line, visualizing perceptions through positioning can reveal a lot of insight. How does this time backwards segmentation and factorizing that we were doing. Perceptual mapping is a form of factorization, it makes sense to segment the respondents first. Because otherwise if I were to average across a hydrogenous set of respondents I would get a confusing map. Which finally brings me to a summary and wrap up for this module. Module 3, what did we see? Well, we saw the following. There was an intuitive understanding of the kind of questions that customer analytics will try to answer. Remember, who are our customers, what are their preferences, what are their perceptions? Questions of that nature. Two, an exploration of the factorization process in some sense. What does it mean intuitively and what happens when I take a big data matrix, chuck it into the machine and basically let it do the factorizing? Three, segmentation possibilities with metric data and we saw that, right? So we basically saw what happens in the cooking oil example for instance. Or, you know, in the psychographics example. What happens when I use metric data and we see these segments emerge? We also saw segmentation with unstructured data, text data. And we actually saw this insight come out from there. Profiling cluster and interpreting clusters. And finally, joint space mapping of consumer preference, of consumer perceptions, all of it coming through together. With that this session comes to a close. [MUSIC]