So in terms of passive, unobtrusive data collection, there are a variety of ways companies can get data from customers. This is just one set of examples. It could be point of sales data. For media planning, for instance, it could be data about radio, TV, social media, kind of audience engagement. It could be web data, mobile data. What we want to do in this particular module, we will talk about examples of different types of data, and more importantly, also related to the type of questions that people can answer using this type of data. So point of sales data is just one example. In this example, for instance, we look at grocery categories, health and beauty products. This is just one example of the type of data that's available to consumers and firms. The typical data chain has about 80 to 100 consumer package manufacturers, many, many households. And the type of information that's available is quite dramatic in terms of geography, product, time, and the type of marketing variables you can have. You can have different kinds of stores, you can have aggregation in terms of SKUs, you can have aggregation in terms of time. Lots of different kinds of data is available. Now, there are many companies that offer point of sales data. I'll give you examples of the big companies that do so. AC Nielsen is a huge force in getting point of sales data, so that's one example. IRI is yet another company, and the third company is SPINS. Now, these three companies basically give you data of different kinds of point of sales. So AC Nielsen and IRI typically deal with traditional kind of stores that you have in grocery stores, traditional products. SPINS gives you lots of information about organic products. So depending upon the company you're in, depending on the products that you look at, AC Nielsen, IRI, or SPINS might be the ideal company to give you this type of data. Now why do people pay so much for point of sales data? First, it's quite complete. You can link aggregate sales to marketing instruments. For example, as a marketing manager, you would like to answer the following question. How do promotions impact my sales? To answer this question, you need good data about promotions, you need good data about sales. Point of sales data can get you that information. It's quite timely. You can get data of different types of aggregation. You can get data within a week, within a month if you'd like to make those kind of decisions. And finally, it's very accurate. Again, going back to the questions that you can try and answer with this type of data, one is impact of promotions. Who buys our products on promotions? Are customers borrowing from future sales? If I, as a customer, buy two cartons of orange juice today, is it borrowing from my future sales? I might hold off on buying future orange juice. Will cherry pickers become loyal? So if I decide, for example, to buy Tropicana today instead of Minute Maid because Tropicana was offering a promotion, will I become loyal to Tropicana? Those are important questions for brand managers to understand. Why? Because that tells you what's the impact a promotion is going to be, not just in the short term, but in the long term. You can also start thinking about the impact of displays, for example. So those of you who go around supermarket aisles, you can start looking, for example, at end of aisle of displays, middle of aisle displays. These are little displays which allow you to see what brands are on promotion. How do they work? Is middle of aisle display better then end of aisle display? It requires concrete data. And finally, you can start thinking about not just what's the action within a particular product category, you can start thinking about action across product categories. And this is very important from a retail store manager's perspective. How is the overall chain doing? What products are selling together? What products are cannibalizing each other? All those questions can be very well answered through point of sales data. What are some caveats, though? One caveat is it misses out on things such as convenience stores. Again, in the grocery products, brands such as Whole Foods, Aldi, Trader Joe's are not there in the dataset. Think back to when we talked about InfoScout. InfoScout is a solution to this. InfoScout actually wants people to take a snapshot of their receipts and send it back to them. So if these customers are buying products at Whole Foods, InfoScout will have that information, but point of sales data may not have it. Another example is you cannot make causal statements. What's the problem here? I would like to know, for example, how does my promotion impact consumer behavior? What you can get a lot of information on is on a promotion that was there. You can get a lot of information on what people are buying, but was there a causal impact of this? We'll come back to this when we talk about causal attribution. And we don't know customer psychographics, and finally, we also don't know what are the exact set of products that consumers were looking at. But still, keeping all these caveats in mind, point of sales data is hugely important data if you start thinking about trying to capture what a customer is buying, where are they buying, what is overall in their basket, and so on and so forth.