I imagine that the new research on the effectiveness of personalized retargeting from MIT Sloan School of Management Prof. Catherine Tucker and London Business School Prof. Anja Lambrecht will be widely discussed for months to come. As MediaPost's Tanya Irwin reported yesterday, the new research shows that retargeting users with highly specific and personalized product information after they leave a retail site may be counterproductive.
I won't reiterate Tanya's reporting here. But in a nutshell, Professors Lambrecht and Tucker used data from Havas Digital involving one of their travel site clients. Using data from the multiple ad networks involved in retargeting efforts, they found that retargeting a travel site visitor with highly specific product information was less likely to lead to a product purchase that day than showing them a more generic travel ad. The more specific product ad did become more effective after that user has visited a review site and gotten more detailed evaluative information about the product.
Because aspects of the study are sure to be discussed and debated for some time, it may be worthwhile to drill into some of the specifics of the scholarly article itself. The working paper is available here.
Tucker and Lambrecht essentially argue that retargeted ads need to match messaging with "mindset." "We suggest that the ability to target consumers with messages specific to their prior behavior means that firms now have to consider whether the design of their ad fits with the mindset of the consumer at that point in time. A mismatch between a consumer's mindset and the type of advertising message could potentially result in ineffective advertising messages."
They distinguish between "concrete" and "abstract" mindsets. In this case the more general travel browser who has not done a great deal of research on a trip will tend to ignore specific appeals even to destinations they may have tried out initially at a travel site. It is only after the consumer has done more research, via a reviews site in this test, that the specific ads become more visible and actionable. In some sense this is common sense. The "abstract" mindset is just another way of saying that the shopper is still at the top of the purchase funnel and simply isn't ready for the detailed offer.
We can only speculate at this point about the consumer process here. But it seems likely that a more abstract consumer is entertaining the general idea of travel and is more likely to be pushed along the decision process by a general appeal.
Some of the results Lambrecht and Tucker show come as a result of the patterns of browsing associated with different ad exposures. For instance, they showed that for the retargeted audience, overall exposure to a contextual travel ad actually led to more conversion on the day they saw the ad than did a behavioral ad or a retargeted ad. This result comes largely because of context. The contextual ad would have been seen on a travel section or site, indicating a user who is actively researching travel. The behavioral and retargeted ads more likely are coming to the user out of context. As the authors admit, "this correlation may simply reflect that consumers who are browsing travel content are more likely to purchase travel products in general. By contrast, the retargeted ads were shown on websites that have content unrelated to travel.
The real shifts in responsiveness to different types of ads come after users went to product review sites like CruiseCritic or TripAdvisor. The probability of converting users with a generic ad before they performed product evaluations was more than twice the rate on specific ads. Before evaluation, generic ads converted at .02% and specific ads converted at just under .01%. After the user visited a product evaluation site, the conversions shifted, but the gap was not dramatic -- about .0125% for generic ads and .015% for specific ads.
There is another variable in all of this:"involvement." The researchers also looked at the relative effectiveness of generic and specific travel ads for people showing different browsing behaviors. As a proxy for indicating involvement in the travel category on a specific day, they distinguished between users who were browsing other travel sites and those who weren't. Even after viewing product evaluations, the less-travel-involved users converted better off of generic travel ads than they did specific ads. But it is in the audience of most involved travel browsers that the effect of the specific retargeted ads really took off, converting at twice the rate (.04%) of generic. But again, involvement seems to be a key determinant of mindset. Those who were browsing travel generally even before going to product review sites specifically, were converting at some of the highest rates of all on both generic and specific ads. It is important to keep this additional variable in mind. As the researchers point out in their paper, "This result indicates that even after consumers approach their purchase decision with a concrete mindset, high information specificity advertising is effective only when consumers are highly involved."
Again, thinking through this research in greater detail confirms commonsensical models. People farther down the purchase funnel or more fully engaged in the process of going down that funnel, will register specific offers being made to them. But Tucker and Lambrecht move us away from generalization about the effectiveness of personalization toward another level of complexity in targeting. The increased coverage and complexity of online ad networks makes it possible for marketers to understand the clickstream of customers and target (retarget) according to mindset and involvement.