Revenue. Finally, the last method that is of interest to marketers is conjoint analysis. Conjoint analysis is concerned with a measurement of consumer preferences. The goal of a conjoint analysis is to measure how a customer values various product attributes. So what you do is that you present customers or respondents with various scenarios through a factorial design. And these scenarios could be, for example, different product designs. Once you have the data, you perform a conjoint analysis to decompose individual responses through a set of tradeoffs in order to measure the utility that customers have for different features. Here's an example for product design. You present customers with different possible product designs for your product. And ask them to pick one option over another such that by the end of the day, they're revealing the mental tradeoff they are forming in their head from one option or another. These tradeoffs could be, for example, be on the speed of the processor, the size of the hard drive, RAM, or the price. By the end of the day, you would have a 16 by 16 factorial design that could help you understand the value that customers have for each of these attributes. Once you have those responses, you are going to estimate customer values for the various product characteristics, which could help you then predict customer choices, for example. The applications of conjoint analysis, however, are larger than that. Let me give you three examples. One is for product design. When you design a new product, you want to know what makes people buy different products. You don't want to have too many products because on the one end that's very expensive, on the other end people may not value all these attributes. Now the question is which features do you want to add to your product, and how do you present those features to the customers? Do you want to have increments of 100 GB for the RAM or for the hard drive? Or is it better to have increments of 500 GB? And so these questions are very important for designing the product and the offering that you have for your technologies, for example. Another example would be designing an advertising strategy. When you design an advertising strategy, there are usually two important decisions that are made. One pertains to media planning, where, and when, and how you're going to advertise. The other question has to do with the message that you want to convey in your commercial. What kind of features, what kind of valuable propositions do I want to convey? Do I want to use humor or not? What is going to make people more responsive to my ad? To do that, you can imagine that you have different characteristics, different options that exist to design your ad. And so, the advertisement here is going to be like the product. Which set of attributes or characteristics do I want to include in my commercial to make people more responsive to the commercial? Finally, one important application of conjoint analysis is pricing. If you manage to infer the monetary value that customers have for different product attributes, then that can inform how you would price your product. Imagine a car, for example. How much more would you be willing to pay to have leather seats in your car compared to standard seats? How much more would you pay to have a sun roof on your car compared to not having a sun roof? So two questions for us. One is, should I include or not these features in my product? The other one is how should I monetize those features? What kind of price or value can I extract from having these features in my product? And so, conjoint analysis allows you to do both. One is to decide which features you want to add to a product or not have for your product, the other decision is how you should price those features. Last but not the least, many people make the mistake to think that conjoint analysis could be used to predict market shares, and that's incorrect. So you would be ill-advised to forecast market shares based on conjoint analysis because conjoint analysis helps you to understand only choices. It doesn't include, for example, how competitors would react to your decisions. And so that's why you cannot use conjoint analysis to make inferences on market shares. Thank you very much.
