Online publishers capture amazing amounts of data about their readers. As a student of consumer behavior, I see a great opportunity to use the data in a novel way: try to identify and categorize patterns of behaviors, and then use a combination of performance data and possibly some neuromarketing experiments to figure out when and where readers are most likely to be receptive to advertising.
Before half of you jump up to say that, duh, this is already done and it’s called behavioral targeting, let me clarify the difference in what I am proposing. To my knowledge, the majority of behavioral targeting is data-driven, meaning that it is based on slicing and dicing data to look for patterns. I am proposing a model-driven approach, which begins with a principled description of the moment-by-moment behaviors of readers, and uses this information to guide data collection and analysis.
Consider the following example: Reader R is in his office and has a short window of time to clean up his inbox. As he skims some newsletters to which he subscribes, he clicks a link to an article whose title he finds interesting, which takes him to website W. Later that morning, he is sitting at the dentist’s office, bored out of his mind as he waits to be called in. He has already read every issue of Field & Stream and Car and Driver, and doesn’t care for Better Homes & Gardens, so he whips out his smartphone and surfs over to website W just to see if anything else interesting catches his eye.
It is my opinion that Reader R is more likely to be receptive to ads in the second scenario, but most data-driven approaches would miss that. A behavior-based analysis of this situation might suggest that if you are visiting a website by clicking a link in an email, your behavior is very different than if you check out the URL in the browser.
Let’s take a more focused example: When everyone was obsessing about placing ads “above the fold,” Outbrain and Taboola realized that a reader who has just finished an entire article is probably highly receptive to suggestions about something else to read. Of course, they proceeded to screw things up by focusing on click rates, which led to the promotion of click-bait titles, which is partly responsible for the sad state of affairs we see today – but that’s another story. The point is that they showed that the bottom of an article is a good place and a good time to present an ad.
Back to my suggestion. A few smart folks with some knowledge of behavioral science or cognitive science should be able to come up a taxonomy of reader behaviors based on real-life situations. It should be possible to come up with fairly broad behavior categories that capture typical sequences of actions, along with their context, whether or not they are goal-directed, and so on.
And then these “models” of reader behavior could be tested through neuromarketing experiments that probe the degree of attention and emotional reaction to ads presented during various key behaviors, and in some cases directly from campaign data.
My suspicion is that a disciplined effort in this direction could lead to some quick results, followed by a growing body of extremely valuable information about what makes readers tick, and when they might be willing — or even eager — to process promotional information. And if done carefully, many of the behaviors could be universal, in the sense that they should apply across multiple Web sites. This suggests that advertisers or even agencies might want to partner with publishers to conduct this kind of research.
Of course if this approach were successful, it would then be up to the publishers to show some restraint, focusing their advertising to the times and places when readers are least likely to get annoyed. And I am not so optimistic about that...