Video RTB Misses The Mark Over Half The Time

Clearstream, a platform that helps advertisers increase online viewability of their videos, released data this week showing wide gaps in the impressions advertisers think they’re buying and the impression that actually gets served. The company expected to see some problems, but the rates at which this type of misalignment was happening far exceeded their expectations, and seems to be getting worse.

The study looked at over 3 billion RTB auctions from 18 brands in September, October and November 2014. In September, 44 percent of the impressions won were not as declared. In October, that rate rose to 58 percent and by November it had reached 63 percent.

“When you win the auction, you don’t necessarily get what you think,” said Brian Mandelbaum, founder and CEO of Clearstream. Mandelbaum says they saw discrepancies across the spectrum from player size to geotargeting, and most prevalently, at the domain level. In some cases no data about the session is delivered to the advertiser at all.

He expects the rates to go down at least temporarily after the holiday retail season when the market is trying to unload inventory. “It’s astonishing for brands to see that they are potential being ‘baited and switched’ over half the time when it comes to RTB.”

Clearstream’s “Post Programmatic Decisions” technology checks for a number of viewability requirements set buy the advertiser after an impression is won, but before it is actually served. If the session criteria isn’t met, Clearstream will either fail the impression or send it back to the SSP to try again.

The rate of misalignment isn’t necessarily all driven by malicious fraud, explains Mandelbaum, but puts a spotlight on a broader issue of inefficiencies in the ecosystem — SSPs, exchanges, yield optimization platforms, publishers — where impressions are treated like hot potatoes.

The ANA recently published a ground-breaking study on bot fraud that found 23 percent of video ad impressions are triggered by bots. Clearstream’s study shows that the issues extend beyond bots.  “There are not perfect standards in the way video inventory can be classified and the way session criteria gets passed through from the advertiser,” he says. Without those standards in place, in effect, you have advertisers and suppliers speaking different languages.