Behavioral Targeting Is Starting to Look Better

The first eye-tracking study comparing behaviorally targeted online advertising to contextual targeting online advertising was released last month. It was conducted by behavioral targeting and media company TACODA with Next Century Media.

The research finds, on average, the same advertising received 17 percent more “looks” on first exposure when targeted to behaviors than when targeted to content. The figure jumped to 54 percent after a second exposure.

The results analysis concludes that though contextual targeting is good for launching new messages, behavioral targeting is more effective at sustaining campaigns.


The study’s hypothesis is sometimes too many ads for the same product category seek the user’s eye in contextual targeting. This causes users to avoid looking at all of them. Even if users were on any given Web site researching a purchase, they might stay focused on editorial content during process, resisting the profusion of similar ad messages. The study refers to this as a “clamor” effect.

On the other hand, when they are seeking a product or service and they find an ad for that product in a completely unrelated context, users may react to that unexpected event by looking at the ad. That’s called a “surprise” effect.


The study’s sponsors engaged The PreTesting Company, which provides eye-tracking services for the advertising and marketing industries, to conduct the field work. The sample was constructed by intercepting 18- to 64-year-olds shopping in New Jersey and California malls for plasma TVs, new cars, and computers.

Two browsing experiences were provided once a subject qualified. One showed behaviorally targeted advertising in any context; the other showed ads only in contextually relevant environments. Four exposures to each of three advertisers were delivered.

The number of times an ad was looked at (looks) and aggregate time spent looking at each ad (seconds) were measured by the eye-tracking tool.


The results show on first exposure, behaviorally targeted ads had 17 percent more “looks” and about the same amount of time spent on average across the three advertisers. On second exposure, “looks” jumped to 54 percent.

By my calucations, on average behavioral targeted ads generate 44 percent more looks than contextually targeted ads.

Behavioral Targeting Contextual Targeting
Frequency Looks Time
Looks Time
1 0.61 1.35 0.52 1.36
2 0.77 1.74 0.50 1.43
3 0.62 1.16 0.38 1.48
4 0.60 1.41 0.41 0.93
Average 0.65 1.42 0.45 1.30
Source: TACODA and Next Century Media, 2006


Before considering the research’s implications, you should keep some limitations in mind. First, the very small sample size precludes this research from being generalize-able for everyone, everywhere. Only 15 subjects were tested for each targeting type.

Second, this eye-tracking study measures looks and time spent. It doesn’t consider any of the economic or attitudinal responses to advertising many advertisers hold up as objectives.

Finally, the current online media planning approach has been built over years of experience. Though results directionally indicate certain shifts could be made to optimize targeted impressions delivery, more significant testing would be needed to break the paradigm.


Disclaimer made, this study’s results have some interesting implications.

The more expected results are the hypotheses about advertising’s clamor and surprise effects aren’t disproved. Most advertising professionals wouldn’t have expected anything different.

Quantitative results showing attentiveness declines with increased frequency of contextual targeting isn’t entirely new, either. The wallpaper concept comes to mind.

However, the fact that this doesn’t hold true for behavioral targeting is somewhat surprising. The report’s authors posit that dealing with this issue by flighting contextually targeted advertising first (I assume to build reach, but this wasn’t explicitly stated), then following up with sustained efforts targeting behavior is a fairly new and interesting proposition.

Final Thoughts

This new report sheds light on some interesting new facts and thoughts about behavioral targeting. It adds to our current knowledge about how this means targeting advertising works and provides a new, independent line of evidence.

It does have limitations and should only be taken directionally, but new thoughts about flighting behaviorally targeted advertising separately are quite interesting. I don’t see how further testing can hurt advertisers, so long as it doesn’t take messages out of the market.

The full study is available as a PDF download, if you’re interested in learning more.

Related reading