So, so far we've talked about Exploratory data collection. We've talked about Descriptive data collection. Next, we want to move on to Causal data collection. As you notice, this type of data collection is more stringent. Let me give you an example. Many times when you're thinking about our websites, of our companies, you think about the landing pages. Think about is landing page A, better than landing page B? How do we determine that? That's where you might want to do a field experiment. Some set of customers see landing page A. Another set of customers see landing page B. We look at the click through rates, and then decide which landing page is better. In other words, we are trying to make causal link between changing the landing page, and looking at the click through rate. That's what causality is all about and that's what field experiments are all about. So there are many companies that allow you to do this type of A B testing. For example, we start thinking about A B testing you need to start distinguishing between correlation and causation. Correlation is the relationship between two variables. So let's take price and sales. Are sales and prices correlated, probably so. If prices go down, sales might drop as well. Causation is one variable producing an effect in the other. Correlation and causation are not the same thing. Let's take some concrete examples to try and understand the difference between correlation and causation. For us correlation and causation there are three requirements if you start thinking about causal inference. First of all is there a correlation at all? That's basically evidence of association between X and Y. Let's say X is price, Y is sales, is there a relationship between them? But that's not enough for causation. There are two other requirements as well. One is temporal antecedents, X must happen before Y. So if you want to say price changes drives sales, we need to be able to say that price changes are good before sales changes happened. And finally there is no third factor driving both of them, In other words, you need to control for other possible factors. So let's take an example to try and understand the relationship between the two and what's the difference between the two. So there's a popular story that storks, for example, bring babies. It typically started in the colder countries where they used to see, for example, a bunch of storks going through, and suddenly landing on a house, and you know a few months later you see a baby in the house. Well, are storks really bringing babies? Well let's try to understand each of the different elements and see whether there is causation going on at all, okay? If you start thinking about correlation, well there might be some kind of correlation. You see houses on which storks are sitting. You see babies in that house. Are they related? They are related. But is it causation? That's the key question. For the one hand, there is correlation. Storks and babies in the same house. Is there temporal antecedent? That's true as well. Storks come in, babies come in a little bit later. But is there a third factor driving both? And that's where, for example, causation and correlation comes in differently. What's the key question? Typically we start thinking about why our storks sitting at a particular house. Typically the answer might be, and this is in colder countries, typically the answer might be because those houses are warmer. Then you start thinking about why are those houses warmer? Typically those houses might be warmer because they have pregnant women in them. in other words, there is a third factor. Houses being reasonable warm that means the stork is coming and sit on top of that house and see a baby coming a few months later in that house. It's about correlation but definitely not about causation. So that's the key difference when you start thinking about correlation versus causation. Now, when you start thinking about the overall story, you start also thinking about many other companies that can help you distinguish between the two. The key idea to distinguish between these two is to start doing field experiments. In other words, you start systematically manipulating prices. You start systematically manipulating landing pages to see how they might have a causal impact. So there are many companies out there that will help you do this AB testing. So a simple example might be, if you start thinking about two pages, landing page A versus landing page B, how do you distinguish which page is better? Well, here's what, ideally, you would like to do. One set of customers coming to your website, will see landing page A. Another set of customers, randomly chosen, will see another landing page, landing page B. Then over time, you see the click through rate. You see the purchases. That can help you determine which landing page is better. This is precisely what many companies help you do. For example Optimizely is one company that will help you this AB testing, it'll help you understand which landing page is better, in other words, monetize overall landing pages and figure out which page you should be showing your customers. There are many, many other companies as well which do the same thing. Overall, when you look at, at this particular space, many companies are starting to help out other companies in terms of landing page design, in terms of overall optimization of the webpage, and so on so forth. So there are many companies that help you do mobile AB testing. What's the idea here? Basically, on a mobile site, you also want to answer what landing pages are better, which icon should be shown, and so on. So a company like Optimizely will help you do that. And it also has many of the companies that might do the same thing. For example, Leanplum is another company that does the same thing, and there are many, many other competitors as well. So when you start thinking about the pricing plan of these different companies, the starter plans typically are free. So you can do a lot of A B testing on its own. But if you start thinking about your overall landing pages, you start thinking about the overall funnel of purchase, that's when you start paying the big money which is the enterprise package. What are the kinds of questions that can help, that can be answered using this type of data? First one is, website optimization. What kind of websites? What landing pages, what icons should shown, and so on. How should they be optimized? Second, you can start thinking about mobile app design. Many companies are coming out with mobile apps. How should they be best designed? And of course, you can take it to the extreme. You can start thinking about how should you design the app in general, but you can also start thinking about what version to the app should be shown to different customers. The extreme case would be one to one marketing. Very broadly, all these companies help you collect data by doing field experiments. In other words, you're systematically manipulating landing pages. You might be manipulating checkout pages and so on, so forth to try and understand how these systematic changes cause changes in customer behavior. So what have we done so far in this module? We've looked at three different kinds of data collection, which go hand in hand with different types of marketing decisions. First, we started off with exploratory data collection, where we talked about focus groups, we talked about Internet communities. Then we went into descriptive type of data selection. We talked about surveys, we talked about point of sales data, we talked about data related to media planning and so on, so forth. And then we ended up with talking about causality. In other words, we talked about field experiments and other company which help you with field experiments. So when you start thinking about the overall summary of this module, it's about data. But, more importantly, it's a synergy between the type of questions you need to answer, and the type of data that's best suited to answer it. It's very, very important to think about the synergy as you start thinking about the type of data you would like to get. If the question you'd like to answer is more exploratory in nature, you do focus groups, you can do internet communities. If the type of question you want to answer is more descriptive, in other words, you want to get concrete numbers in terms of share of wallet, in terms of share of voice, then the type of data you need to collect is quite different from the exploratory type of data. And finally, if the question that you need to answer is more causal in nature. For example, if I changed my landing page, what would happen? Then the type of data you need to collect would be more along the lines of field experiments. So it's very important to keep the managerial goal and the the type of data collection in mind because they're highly synergistic.
