The Dirty Little Secret of View-Through Conversions

The measure of "view-through conversions" (VBC) have been controversial since their invention. They capture instances of a user seeing the advertiser's ad, not clicking, but subsequently visiting, and registering or purchasing from the advertiser's site. Some consider VBCs a measure of direct response and use them in their ROI calculations, while others believe they measure happen-stance events and disregard them completely.

The fact is, it's difficult to justify VBC as a measure of direct response. In order to do so, you'd have to prove that those impressions that were not clicked had a significant causal relationship to the subsequent sale, registration, or visit to the site. Pessimists correctly point out that if you put a banner in your rotation that has nothing to do with the advertiser, that banner will garner some view-through conversions. In an attempt to conduct cleaner experiments, some have conducted "test/control" (a.k.a. "True Lift") experiments. In almost all cases, both the test and control groups exhibited the same natural tendencies to visit the advertiser's website and convert. Or in other words, True Lift tests do not support the notion that VBCs are a measure of direct response. This is especially true for advertisers with strong brands. The more popular the advertiser's website, the harder it is to measure significant lift because of all the other factors that contribute to consumer activity (e.g. offline advertising, previous visits, word-of-mouth, in-store history, etc.).

So, if VBCs aren't a measure of direct response, what are they? VBCs are a measure of targeted reach, not direct response. Let me use an analogy to illustrate. Imagine you were responsible for Coca Cola's advertising, and you could ask every person who just bought some Coke whether they had seen a Coca Cola ad recently. And if they had, they had to hand over a very accurate log book of all the Coca Cola ads they were exposed over the last 90 days. Furthermore, the notes would tell you which specific ads were delivered, and the context of where the ads were displayed. For all those people who answer "yes", do you think it was that last ad impression that caused the person to buy that Coke? Of course not. But would that information would be valuable to Coca Cola? Absolutely. With it, they could measure where their customers are being reached by their ads and how often.

View-through conversions provide very similar insight. Imagine you reached 1MM users on Site-A and 1MM users on Site-B. And you get 200 view-through conversions from Site A and 100 view-through conversions from Site B. What would that mean? It's a strong indicator that Site A has about twice the concentration of your target audience than Site B. That may be interesting for the average direct response advertiser, but amazingly profound to the typical brand advertiser.

Though there isn't a strong argument for VBC as a measure of direct response, there is a strong case for VBC as a measure of targeted reach. Unfortunately, the metric has the word "Conversion" in it rather than "Reach", a hold-over from the early days when the only justification for the web was direct response. As the industry matures, and we create compelling reasons for brand advertisers to invest in web advertising, the repositioning of these metrics will become supremely important. It's time to drop the ROI moniker where it isn't sincere, and start saying with a straight face what things really are. Scary, I know. but trust me. good things will happen.

Thanks

Young-Bean Song

More from Microsoft Atlas Institute