|
||
EyeReturn‘s Ian Hewetson, vice-president of client services, has dealt with a lot of fraud.
Speaking at the IAB Spring Mixx conference Thursday, he recalled when his team found clear markers of fraudulent traffic on one publisher’s site. The publisher was in the U.S. and was someone Hewetson saw as legitimate — so he called him and shared his suspicions.
“There’s no fraudulent traffic coming to my site,” Hewetson remembers the publisher saying. “Everything’s totally fine.” But Hewetson sent him some of EyeReturn’s traffic data. The publisher agreed it looked like fraud, but said he hadn’t bought any cheap, obviously fake traffic. It had to be one of the networks he dealt with, pushing traffic to his site to bump their own numbers. Hewetson believed him; he seemed genuine.
But then Hewetson asked, “What are you going to do about this? Are you going start investigating the SSPs, maybe start chopping them off one by one until you find out the source of it?”
“Why would I?” the publisher replied. “It’s nothing to do with me — if you hadn’t come to us, I wouldn’t even have known about it.”
Hewetson says it’s that kind of apathy that drives fraud in the digital ad industry. Not every publisher who gets fake traffic is deliberately trying to defraud advertisers, and many aren’t even aware it’s happening. But if they do find out, there’s no real reason for them to stop it. After all, their balance sheet now depends on that traffic. Cutting out the fraud could mean cutting the site’s revenue in half.
Other players along the supply chain aren’t motivated to pay attention, either. Sites that get fake traffic are often legitimate, or at least look that way at first glance. That’s enough for some supplier networks, exchanges, and even agency traders — who all take a cut of fake impressions bought on those sites — to turn a blind eye.
“It’s an insidious problem, because there’s not much incentive to stop it,” Hewetson says. “Everyone’s profiting off it. The only party in this entire transaction that’s losing is the end advertiser.”
And the advertisers — who do have a clear incentive to stop fraud — don’t have the power to catch it. The surest way to catch a bot is to track its journey across sites. A typical human user will go from a story about the missing Malaysian plane to a Wikipedia page about historical plane crashes; but a bot user will visit 100 other sites with strikingly similar content. It’s a big red flag — but you have to be able to follow the user across sites.
Fraud investigators need to have wide reach, Hewetson says. They need to be able to follow potential bots from site to site, picking up traffic patterns that look suspicious. Once the bots are identified, they can be quarantined, along with the sites they visited, so clients won’t purchase any more impressions. EyeReturn sometimes approaches publishers or networks where it finds fraud, but Hewetson says that usually goes nowhere. A better strategy is to cut the supplier off, to show they mean business.
Once a DSP detects fraud and cuts it out of their network, that’s usually the end of the story. Although Hewetson didn’t bring this up in his talk, players that detect fraud don’t have much incentive to share that hard-won information with their competitors. Large botnet reveals are big news, but how often does Google DoubleClick cross-reference its blacklists with AppNexus? Without an industry-wide network for fraud reporting, a bad publisher shut out of one exchange or DSP can keep on filching from all the others.
Although Hewetson says it’s big suppliers and exchanges that are ultimately responsible for controlling fraud, advertisers can’t just wait for it to go away. One thing he recommends is for every digital buyer to avoid click and CTR optimization, and use performance indicators that are harder to simulate, like post-click pageviews on the advertiser’s site and purchase conversions. The more fraudsters have to customize their bots to game an advertiser’s algorithms, the harder it will be for them to get the scale they’re looking for. That makes a campaign a far less appetizing target.