For a long time, I have argued that online users have a sense memory for publishers that abuse their attention. I think we reserve a special bit of grey matter for the sites that give us trouble. On some level, we recall that site X loads slowly because ad or design bloat drags down its page loads. We record somewhere in our brains that site Y is too reliant on intrusive pop-ups and full-page interstitials. And we know that site Z slaps a 30-second pre-roll ad onto a 90-second clip. Without a whiff of proof beyond my own surfing experience, I contend that many of us curtail our use of these sites, even if it is by a couple of visits a month or a few video clips not launched. Sites do an immeasurable level of damage to themselves by stretching the bounds of visitor tolerance.
On some level, most publishers know this too. Most of them do their own cost-benefit analysis of how annoying, ugly, slow, or unfair their ad loads must be before suffering diminishing returns from all of those desperately needed ad units. It turns out, they may be wrong on the math.
Some researchers have been working to prove not only that the quality and annoyance of advertising can negatively impact site use, but they also are trying to quantify the actual cost of "badvertising" to publishers. In a late 2014 issue of the Journal of Marketing Research, a study by analysts at Microsoft Research and Northwestern University ("The Economic and Cognitive costs of Annoying Display Advertisements") contends that lousy ads may cost sites more than they make from the ads in diminished page views.
They started by assessing how users perceived ads and creating a pool of units that fell above and below a mean of annoyance. It probably comes as no surprise that the ads designed to attract or derail attention were deemed by users to be the most annoying. But it's worth noting that animated ads were almost always rated more annoying than static ones. In fact, even in their written explanations in the survey, respondents most often referred to movement in the ad as the source of their annoyance. The other most-cited annoyance was just plain ugly execution.
But things got really interesting once the researchers built a pool of more and less annoying ads. Using the Amazon Mechanical Turk workforce, they paid people to review and classify email pages with different types of good and bad ads and pages with no ads at all. They then compared how many page impressions users generated from these three different conditions at three different pay scales for doing the tasks before dropping out.
The results were striking. Even when being paid to view pages, users were considerably less willing to continue the task after viewing pages that had ads deemed annoying. The researchers contend that the cost per mille (CPM) of bad ads (versus no ads on a page) in this test was $1.53. Considering that much of the worst advertising online generates considerably less than $1.53 CPM, they argue that the cost of badvertsing is often higher than the bottom-line benefit. It is also worth noting that pages with less-annoying ads had very little impact on the user's willingness to engage the task and caused little dropout.
But bad ads may affect more than the bottom line. The researchers also explored how annoying ads affected the reading process. Publishers may be heartened by the finding that annoying ads actually increased the amount of time users spent on a page. Using mouse tracking as a proxy for attention, they found that bad ads successfully distract visitors. "Distraction" is the key word here, because bad ads also seem to diminish overall concentration. Even though users actually read the text on bad-ad pages in the same way they read it on good-ad and no-ad pages, their comprehension of the article is less accurate.
This research, while far from conclusive, underscores the sad reality of 2 decades of ad-supported content. We have created one of the most annoying media consumption platforms in human history. Clogged with bad ads that often bear no visual or contextual relevance to the content nearby, most sites eschew what legacy media always coveted: crafting desirable media environments that serve users and advertisers.