No one is certain how badly digital inventory is exposed to fraud, and end-of-year data from influential fraud trackers doesn’t make the picture any clearer. DSP Solve Media, author of the quarterly fraud-tracking Bot Traffic Advisory, reports that suspicious ad traffic in the U.S. climbed steadily through 2013, culminating in a 60% jump from third to final quarter. Ad verification provider Integral Ad Science reports the exact opposite — that suspicious traffic has fallen to its lowest level yet.
First the bad news. In its first year-end Bot Traffic Advisory, Solve Media claims suspicious U.S. display ad traffic grew 40%, while suspicious mobile traffic grew 30%. Through the holiday season, Solve marked almost two thirds of all ad traffic as suspicious, while one quarter of mobile traffic was suspicious.
“With the holiday season comes an increase in advertising spend. It’s not surprising that there was a spike in suspicious traffic in Q4,” CEO Ari Jacoby wrote in a blog post. “With US advertising budgets expected to top $182B by 2015, brands and publishers must both commit to adopting fraud prevention measures now.”
Solve’s results, culled from 700 million ad verifications across 8,400 publishers, have typically been the most alarmist among verification providers’ estimates, which average around 20-30%. Solve’s estimates may appear high because its monitoring focuses on display advertising, and because it reports all suspicious traffic, rather than confirmed fraudulent traffic. On the other hand, by offering less conservative estimates, Solve may be highlighting a much larger fraud problem than anyone suspects.
Integral Ad Science’s outlook was much more optimistic. It reports fraudulent U.S. traffic on exchanges decreased 57% in 2013, from 30% of exchange inventory in Q4 2012 to 13% in Q4 2013, and credits growing awareness and understanding of the fraud threat across the industry. “Now that both advertisers and publishers are taking measures to protect themselves and optimize their media choices, overall quality has improved,” said CEO Scott Knoll in a release. “We’re seeing more ads are in view, and brand safety and suspicious activity problems are being tamed over time.”
As part of its report, Integral released a bot traffic study that investigated the habits and concentration of bots across the U.S. It found nonhuman traffic made up the largest chunk of ad impressions in the early morning, when most legitimate users were asleep, and tracked time zones across the continent. It found rural areas, which have a lower density of users, are susceptible to higher concentration of bot traffic.
Fraud was most prevalent on publisher networks, followed by RTB exchanges. The lowest exposure came from buying direct from publishers, which sees a consistent fraud volume of around 2%.
Integral’s data comes from 2B daily impression verifications through its integrations with major DSPs, exchanges, agency holding companies and advertisers. An estimate of 13% fraud volume puts it at the low end of verification providers’ estimates.
Conflicting data on the prevalence of fraud can give advertisers and media companies a headache, especially when attempting to budget for expected losses. But although there is disagreement over the extent of the problem, most experts agree on the solution: more awareness about impression fraud, including how it gets into the supply chain and how it impacts publishers and advertisers.
Currently, fraud trackers do not post data on the Canadian market.
By the numbers
Solve Media suspicious traffic estimates, 2013
• Q1: 44%
• Q2: 42%
• Q3: 51%
• Q4: 61%
Integral Ad Science suspicious traffic estimates
• H2 2012: 30%
• H1 2013: 20%
• H2 2013: 13%