February 9, 2016
Advertisers will lose an estimated $7.2 billion dollars in 2016 to fraudulent impressions created by bots, according to a new study by the Association of National Advertisers (ANA) and cyber-security firm White Ops released in January.
The study tracked the digital advertising for 49 companies over a 61-day period in the late-summer and early-fall of 2015, and concluded that aggregate bot frauds levels had remained roughly the same as they were a year ago. Though, less than one-third of advertisers surveyed in both 2014 and 2015 saw decreased bot rates.
“’Fraudsters are hidden in the very complex digital advertising ecosystem,” says Bill Duggan, group executive vice president at the ANA, in an e-mail. “The fraud perpetrators don’t have a human face and that makes it difficult to catch them.”
Michael Tiffany, CEO and co-founder of White Ops, says the companies that did experience a decline in bot rates were among the hardest hit the previous year, and thus inclined to take a more active role in preventing further fraud.
“If you saw double digit fraud percentages last year then you got a wake-up call,” Tiffany says. “So what happened is that each one of those marketers changed their practices to get out ahead of this problem.”
Meanwhile, companies reporting average or below average rates of fraud last year were more likely to see their bot rate increase, which Tiffany ascribes to complacency.
“They often treated that good news as de facto approval that the systems that they had in place were dealing with fraud effectively, so they didn’t change their behavior and fraud actually rose for the majority of those marketers.”
“The lesson to be drawn here is that this kind of fraud and these bot net operations are not effecting every marketer all the time,” he adds. “The 20% of victims that are hit the hardest are paying for 80% of the crime at any one time.”
The survey also takes a look at how bot developers continue to profit from their creations, despite enhanced methods of detection. While cyber criminals create bots with the expectation they will be discovered and destroyed, ad fraud remains lucrative by continuously infecting new machines.
“What we found is that profits to the cyber criminals remain unaffected as long they’re infecting new machines at the same rate that their old infections are getting caught,” Tiffany says.
Others findings in the study include:
• The more expensive the media, the more susceptible it was to bot fraud, owing to the high-value of the target. Display media with a cost per thousand impressions exceeding $10 had a bot rate 39% higher than cheaper display media. Bot rates were 173% higher for video media with cost-per-thousand impressions greater than $15 compared to low cost video media.
• Programmatic ad buys also had higher levels of bot fraud. Fraud occurred 14% more often in programmatic display ads and 73% more often in programmatic video ads compared to average. “The programmatic supply chain is complex and murky. That murkiness lends itself to fraud. For every $1 that an advertiser spends on programmatic advertising the publisher ultimately gets 35 cents. Where does the other 65 cents go? While there are certainly intermediaries that add value there are others that don’t,” Duggan says.
• Bots inhabited sourced traffic in greater numbers and with a higher level of sophistication than unsourced traffic.
• Campaigns targeting specific demographics saw higher level of bots. Hispanic targeting, for example, experienced bot fraud at levels nearly twice as high as non-Hispanic targeted media.
The study also provides recommendations to combat ad fraud, including using third-party traffic monitoring, requesting transparency from sellers about sourced traffic and the programmatic supply chain, and ensuring anti-fraud policies are followed by all external partners.
“I believe the path to success here is fundamentally about disincentivizing the entire supply chain from paying for the bot traffic in the first place,” Tiffany says. “Certainly the technical arms race is a big part of winning this war, but it needs to be structured in such a way that the supply side of this industry faces consequences for sourcing bot traffic in the first place.”
Or, as Duggan say, “The fight is on. Buyers should not ‘accept’ any fraud. With elevated awareness and consideration of anti-fraud policies and strategies, the trend can indeed be reversed.”