• by Mark Walsh  @markfwal, March 26, 2012

Earlier this year, comScore revealed some key findings from a broad study of online brand campaigns in connection with the launch of its ad verification service called Validated Campaign Essentials. Among the highlights, it found that nearly a third (31%) of online display ads are never seen, and 72% of campaigns had at least some ads running next to objectionable content.

Following up, comScore has now released the full results of its December study, which spanned a total of 18 campaigns, nearly 3,000 placements, 380,898 site domains and 1.8 billion ad impressions. Participating marketers included Allstate, Chrysler, Discover, eTrade, Ford, General Mills, HTC, Kellogg's, Kimberly Clark, Kraft Foods and Sprint.

One of the obvious implications of a third of ads not being in view -- defined as at least 50% of an ad’s pixels being in view for at least one second -- is that ad dollars are being wasted. The report found little to no correlation between CPM and value being delivered to the advertiser. Ad placements with strong in-view rates are not getting higher CPMs than placements with low in-view rates.

One of the common misconceptions it points out is that ads “above the fold” are seen, while those “below the fold” are not. comScore found that some ads delivered ‘above the fold’ were not seen because users quickly scrolled past them before the ad had a chance to load, and alternatively, many ads placed below the fold had a high chance of being seen.

This might mean that below-the-fold inventory can be priced as premium as long as the publisher can prove it was viewed. “Even publishers aren’t able to price their own inventory without having misunderstandings,” said Anne Hunter, VP, ad effectiveness at comScore.

Looking at common ad sizes, the study found the classic leaderboard (728 x 90) had the highest in-view rate (74%), followed by the medium rectangle (300 x 250) at 69%, and the wide skyscraper (160 x 600), a bit lower at 66%. But even with the leaderboard, there was wide variation across sites, with the in-view rate ranging from 7% to 93%.

Larger sites within a given content category also tended to do a better job of making ads viewable than long-tail sites, at 77% versus 60%. And among particular content verticals, coupon and directory sites had the highest rates for viewable ads -- at 89% and 81%, respectively. By far, the worst were pet sites, at 27%.

comScore also examined campaigns based on their ability to target by demographic or behavioral factors.

Campaigns that included one demographic variable (such as 25-54 years old) successfully hit that target audience an average of 70% of the time. In cases where there were two variables (age + gender), the accuracy of targeting decreased to an average of 48%. And with three variables, the average was dropped all the way to 11%. In effect, the narrower the demo target, the harder it is to nail. 

As with ad visibility, comScore found no correlation between CPMs and the accuracy of demo-targeted campaigns. 

When it came to behavioral targeting, the average campaign reached its intended audience 36% of the time, with a range of 23% to 67%. Hunter suggested that the uneven results stem largely from the limitations of cookie-based targeting.

But the study confirmed marketers’ interest in behavioral data. Hunter noted that after a certain age range, marketers most often wanted to target a certain type of behavior -- people spend a lot of time on automotive sites or cooking sites, for example. “So the target delivery against behavioral is just as important as against basic demographics,” she said. 

On a more technical note, comScore noted that all of the ads in the study were delivered via iFrames -- an HTML feature used to wall off ad content from a page’s main content -- and 61% through cross-domain iFrames. In its ad verification guidelines issued last month, the Interactive Advertising Board advised minimizing the use of cross-domain, or “nested” iFrames because of the challenges they present to ad measurement.