Independent oversight is coming to help police YouTube videos for brands, but there will be limits to what these outside parties can deliver.
The ad tech watchdogs won't be able to vet every YouTube video for terrorist tutorials, hate speech and sexual content in time to stop them from ever carrying ads, according to ad tech experts.
"A lot of brand safety vendors are jumping all over this opportunity," said Augustine Fou, an ad fraud researcher. "But they're either over-representing what they can do or just not being truthful about the technology."
As it tries to reverse a sudden mass exodus of major marketers over content concerns, YouTube has announced expanded partnerships with third-party ad tech vendors such as Integral Ad Science and DoubleVerify so they can audit the platform to ensure ads run appropriately. Advertisers began freezing their ad spending on YouTube last month after press reports revealed their ads on problematic videos there. They say they'll return only once they're assured that it won't happen again.
YouTube is missing out on millions of dollars a day and could lose $750 million or more this year if it doesn't turn things around, according to some analyst estimates.
The third-party vendors are just one piece of what YouTube is implementing. It has given more controls to brands to opt out of ads before certain videos. It has also updated policies to make sure its community of creators abides by its standards, being "demonetized" if they don't.
As for the third parties, the details are still being worked out, and the companies don't seem to have a full grasp of what they can accomplish.
"We have been notified by Google that DoubleVerify has been chosen as one of the third party measurement partners to help solve the brand safety issues that have been identified by our joint customers," a DoubleVerify spokeswoman said in a statement. "As of yet, we have not received any details on the solution that they are planning."
The ad tech companies will act more like auditors, reviewing campaigns after the fact, according to people familiar with YouTube's thinking.
The third parties will be able to conduct brand safety reports that show advertisers where campaigns ran. If some small fraction of the ads went to inappropriate video inventory, that could help marketers get off of those videos and avoid having to pay.
Integral Ad Science was hoping for even more control, to stop ads from ever running before bad videos, according to CEO Scott Knoll.
"Details are still getting worked out," Mr. Knoll said. "I don't think anyone's goal is to provide data after the fact. The whole goal is to prevent this from happening."
Mr. Knoll envisions a system where Integral Ad Science could review every YouTube video and assign a risk score. Advertisers would decide what risk threshold works for them.
The concept is not unique to Integral Ad Science. Agency holding company giants WPP and Omnicom each announced last week that they would vet millions of videos and assign risk scores. The media holding companies plan to offer whitelisted YouTube videos to advertisers.
That approach will shrink the pool of inventory because every piece of content has to be pre-approved.
YouTube's strategy appears to be different: It wants to eliminate the bad actors as they arise and focus on getting better at identifying bad actors more quickly. It also wants to make sure marketers keep considering buys across all of its massive inventory, not just picking from someone's pre-approved list.
There are more than 500 hours of content uploaded to YouTube every minute.
YouTube has shown some advertisers that their ads received a handful of ad impressions on truly bad content out of thousands of "safe" impressions. It hopes brands decide that the issue is manageable and that the third parties can reinforce their confidence.
But third parties won't likely be able to deliver much more than that, according to Mr. Fou. If Google can't catch this stuff itself, he said, there's no third party that can do it any better.