Welcome back. In this lesson, we're going to discuss ways of attributing success of a sale to different types of advertising that you might be employing as part of your advertising strategy. By then end of this lesson, you should be able to describe the different ways marketers attribute a successful sale to the media used in their marketing efforts. Now, of course, there is no single way you can do this. Let's return to the scenario we looked at earlier with the purchase funnel, which was awareness, concentration, intent and decision. Consider, for example, an individual who looks at display advertising, social advertising, paid search, organic search and direct marketing and then proceeds to make a purchase on your website. In this case, it's not always easy to ascertain how much you can attribute the success of a completed purchase to the different forms of advertising that you employ. Now, there are several different methods of attributing success to types of advertising that you do, and we'll discuss some of the more commonly used ones. One of the most commonly used methods is a last interaction attribution, which says that the purchase is attributed to the very last ad that the consumer saw. Now, of course, this ignores the long-term effects of ads that appeared earlier in the funnel. As you saw before in the scenario that I had shown, that a consumer might have seen a display ad initially, but now, because you're only attributing all success to the last ad that they saw, you're overrating the most recent ad and underrating the effect that display advertising had. On the other hand, you could have a first touch interaction where it's the first ad a user clicks on or sees that accounts for 100% of the sale. Now, of course, that means that you're ignoring all the other ads that occurred later in the funnel. You could have also a linear attribution model where all ads receive equal weight. Now, while that seems okay, when you think about it at first, remember that you could be overrating low-quality ads and underrating higher-quality ads in your ad portfolio. So again, that's not the best approach for any particular situation either. Time decay is another model that is used. That's when all the ads get similar attribution but more recent ads get a higher rate than the earlier ads. Now, the problem,of course, here is that this is ad-hoc. So how much weight should ads early in the stream have, as compared to ads that are later in the stream? Another model is called positon-based. So with this one, you can rate ads accordingly based on the position in the part of the purchase they had. Now, of course, again, like all the other examples we saw, it's more ad-hoc. Another model is regression-based attribution. This is more scientific, where the rates are attributed based on the responses that you receive directly from the consumers. But of course, you would also, again, be providing significant weight to advertising that may not have played a relevant role in the purchase process. For example, if you're looking at a website that has some relevant text to you and the ad is displayed, but you completely ignored that ad, a regression-based model would still count the fact that you looked at the ad, even though the ad had almost no effect on you. Another method is experiment-based attribution model, a type of A/B testing. In reality, this is probably the most accurate way to determine ad effectiveness, but of course, it is difficult and expensive to conduct. In such a scenario, you would first divide your population of prospective consumers into different types. Next, you would construct multiple types of ads, use multiple media and assign sequences of these media differently to different sets of individuals. You would then study how the different sets of consumers responded differently to the different advertising exposures. Examining these differences will help you identify the best media, and the best sequence of media that generate the most sales. As evident from the subscription, it becomes quite expensive as the number of media and consumers increases. In many cases, you would need several hundred thousand, even a few million customers to get accurate enough results. Now, if you have the budget for this then I would recommend it. An experiment based method to determine addressing effectiveness is a great way to see how advertising directly impacts sales. And so in your market research report, running an A/B test can provide a lot of insight into how advertising and sales might be related. Just remember that these types of tests are quite expensive to do, and so there are other approximations like the ones we discussed earlier. In this lesson, we have looked at different ways the effectiveness of digital media on sales can be measured and attributed. There is no optimal method. Some methods are better and more scientific than others. Given this information, however, it is your firm situation and circumstances that will determine the best method for you.