Ad fraud, the use of computer “bots” to make it seem as if a human has seen an ad, dropped drastically last year, according to a report released this morning by Integral Ad Science.
“Suspicious” ad activity fell from 30% in the fourth quarter of 2012 to just 13% in the just-ended last quarter of 2013, according to the company, previously known as AdSafe Media. The company, which helps advertisers and publishers determine the quality of media, raised $30 million in a venture round earlier this month.
Why the big drop? Greater awareness of the problem and thus more use of technology to reduce fraud, CEO Scott Knoll said in an interview. Integral says it can spot suspected bot activity in real time and remove it before an ad is served.
Still, 13% is a lot. That means almost one in eight ad impressions is fraudulent. ”It’s still at an unacceptable rate,” Knoll says. That said, he thinks that, like click fraud in search advertising, it will never go away because it’s a constant game of cat and mouse. What’s more, advertisers and publishers often turn a blind eye, since each can benefit from fake ad impressions–publishers because they get paid per impression fraudulent or real, and advertisers because they often want to fulfill a certain volume of impressions they want for a particular campaign.
Another metric advertisers and publishers are interested in is whether an ad was viewed. That also has improved, though much less. Some 66% of ads placed placed directly with websites were “in view”–that is, at least half the ad was visible for one second or more–which was five percentage points better than six months ago. About 51% of ads placed through ad networks were viewable, while 47% of ads bought through ad exchanges were viewable.
Again, even if viewability is improving, it’s a long way from ideal. “It’s still not what you’d hope for,” Knoll says. But for one of the same reasons fraud remains persistent–an incentive by most parties to count ads whether they’re seen or not–it’s unlikely viewability will ever get close to complete.
Those two metrics as well as other measures such as the amount of “brand-safe” ad inventory, the professionalism of the website, and the amount of ad clutter on a pages, figure into an overall ad quality score Integral calls TRAQ. Overall, not surprisingly, websites get noticeably better TRAQ scores on average than ad networks and exchanges, some of which continually deal in fraudulent traffic. Websites got a TRAQ score in the fourth quarter of 684, compared with 597 for ad networks and exchanges.
Of course, neither emerges smelling like a rose, since that’s out of a potential top score of 1,000. Knoll said quality varies widely both among websites and among networks.
What’s more, getting further improvements will be much harder, especially when it comes to fraud. “We got the low-hanging fruit,” he says, including “amateurs” taking advantage of limited knowledge and policing of fraud to date. “It’s going to get harder from here.”
Eventually, he says, lower fraud and higher viewability should result in higher online ad prices.