As ad dollars increasingly pour into streaming video, fraudsters are wasting no time in attempting to illicitly siphon off their share of the spoils.
MadHive, an advanced TV advertising company, estimates 20% of streaming video ad requests are fraudulent, up from an estimate of 18% it made in March. Using that percentage figure and other industry estimates for over-the-top (OTT) ad-spending, AdLedger, a nonprofit consortium working on blockchain standards for online advertising, predicts marketers will waste $1.4 billion on fraudulent OTT ads in 2019.
IPG’s Magna Global estimates U.S. OTT ad spend will grow 39% to $3.8 billion in 2019 and will reach $5 billion by 2020.
“OTT is so expensive and for a long time there was scarcity in the amount of inventory out there…in the last year or two, supply [of OTT] has exploded but prices remain high, meaning it’s an attractive target for fraudsters,” said Christiana Cacciapuoti, AdLedger executive director and vp of partnerships and platform operations at MadHive. Still, some in the industry believe there isn’t any fraud on OTT.
“As someone who grew up in digital, I knew that was impossible,” said Cacciapuoti, who previously worked at ad tech firms including Tremor Video (now known as Telaria) and Wochit.
While some fraudsters are merely transferring the skills they picked up from manipulating ad auctions elsewhere on the web to video streaming — from bots to spoofing — there are some idiosyncrasies to ad fraud in the OTT space.
Supply-side platforms (SSPs) can also manipulate ad buys, even if buyers think they’re buying from a premium publishers.
“We’ve seen SSPs and ad networks buy an entire pod, cut it into smaller ad slots, then resell those, give themselves a free ad slot and arbitrage the rest of it,” said AdLedger’s Cacciapuoti. The publisher might not notice what’s going on, unless they get access to reporting from the SSP and know where to look when they do.
Other common types of ad fraud in the OTT space include geography misrepresentation — where fraudsters seeking high CPMs falsely claim their traffic is coming from U.S. IP addresses — and app spoofing, where, much like domain spoofing in display ads, the fraudsters falsely represent their inventory as belonging to premium apps.
Elsewhere, the nature in which OTT ads are grouped together in pods can sometimes lead to false positives. A fraud technique known as “looping” describes when the same device or IP address sends out ad requests than could possibly be viewed within a short period of time. One person couldn’t view five 30-second spots in 60-seconds, for example. Yet, a big chunk of OTT viewing is livestreaming, where the ad server tries to fetch ads before the commercial break for all users watching the stream at the same time, according to DoubleVerify svp product management Roy Rosenfeld.
“It looks exactly like fraud — you really have to go into the deep end and get the right set of methodologies,” said Rosenfeld. “If you flag things like that as fraud, when it’s not, you’ll do more damage than good.” Broadly speaking, DoubleVerify sees OTT fraud rates at anywhere between 2% to as much as 90%, depending on the campaign.
Cacciapuoti said buyers and their fraud detection providers can filter this out by comparing the ads that were fetched with the ads that were ultimately rendered and viewed. Beyond working with third-party fraud detection services, one of the best ways advertisers can protect themselves against OTT fraud is to push their vendors into being as transparent as possible with their reporting, both Cacciapuoti and Rosenfeld advised.
Michael Nevins, chief marketing officer at ad server company Smart, says any new tech creates a “murky environment” that attracts bad actors. Added to that, the increase in demand for OTT inventory and rising CPMs creates a “perfect storm” of opportunity for fraud.
“I’d expect the fraud vendors to become more sophisticated, which will help, but the device industry is still quite fragmented without strong standards yet,” Nevins said. “This presents a challenge.”