For years, we were told online ads should be relevant to our sites. That never really made sense to me, because it often meant advertising your competition. You’ll notice TV shows don’t run commercials for other networks’ TV shows, right? Why should we advertise websites providing the same type of content we provide? And how strange it looked to me as a visitor, when someone ran an ad for a similar site. It was like they were saying, “But my content’s not that great. You should check out this other site!”
Another shortcoming of relevant keyword advertising was that if someone had found the content they were looking for, they would ignore an ad for similar content. But if they were reading about big screen TVs when they came across an ad for a new MP3 player (two products loved by similar demographics), they might decide they’d like to check out MP3 players, too.
Then came behavioral targeting, which meant cookies would follow visitors all around the web, spitting up ads for adult diapers on sites that reviewed happening Miami nightclubs because someone on that computer had recently searched for adult diapers. This creeped visitors out. Even people who understood very little about how the internet worked felt like someone was stalking. Grandparents called their grandkids: had they gotten a virus? And now internet giant Microsoft is going to protect you from it. The idea of targeting the specific visitor and his/her specific wants and needs was good. But there was no good way to execute it. Not only did it scare people away; it also failed on shared computers, where perhaps one family member had been searching for adult diapers, but now someone with no interest in them was surfing.
Could it that somewhere in between the two lies the feasible option?
Sensible Targeting
The TV model – targeting demographics – makes sense. It’s just been hard to implement perfectly because there’s no way for TV advertisers to know precisely who’s watching a given show. The same problem plagues websites. Behavioral targeting let us know where else the user at a particular computer was going, but fell short in all the ways described above. But companies like Quantcast are beginning to do the same for the web as Nielson did for TV: they can give us at least a general idea what sort of visitors like a site. Or, at least, what age, race, income level and so on a visitor is.
But how much does that really tell us about the ads he or she will respond to? It suggests general trends, but sometimes those trends aren’t dominant. For example, what if 55% of men aged 49-65 are interested in widgets? Do you advertise widgets to them, and bore 45% of your target audience, or not?
Working with Personas
Another option is to work with personas. These are imaginary visitors to your website whom you flesh out like fictional characters. Holly Buchanan has a detailed guide here, but in a nutshell: if you already have some demographic info indicating your biggest audience share is middle class women with college degrees, you’ve got a start. Now start asking yourself questions. How do they live? What do they do all day? As this person becomes more and more real to you, narrow in on the important questions: who do they shop for, and what do they buy? Or, if your market is totally about content: what do they read, and who do they share it with?
You’ll have to make some educated guesses. But once you’ve built a persona or two who loves your website, then you target your advertising to them.
Does it work? A lot of smart people think it does. It makes sense. No matter how much science tells you who is coming to your site, it takes some intuition to figure out what they want – because oftentimes, even we don‘t know what we want until we see it.
There won’t be any serious data to prove whether this works or not until advertisers either embrace the concept, or at least provide us with the right tools to test it ourselves.
Leave a Reply