Isis videos may be in the rearview mirror for some advertisers, but what about Peppa Pig?

Brand safety worries have subsided for marketers who've moved back to YouTube despite the lack of ironclad guarantees that their ads will never again appear with terrorist or hate videos. But the controversy has exposed a relevance problem that might be even harder to address -- with ads showing up on children's videos, for example, when kids aren't brands' target.

A spring of discontent
Big advertisers that have come back to Google's YouTube after boycotting over horrible ad adjacencies include Johnson & Johnson and Nestle. Executives of some returning companies said they were more worried about getting tarred by the surrounding news coverage, which has subsided, than consumers seeing the unwelcome placements on their own.

Other major players including Walmart and Procter & Gamble continue to avoid YouTube, meanwhile, pending stronger promises that their ads won't appear with questionable content. (Walmart confirmed that it was still out; P&G declined to comment.)

And some, like Unilever, never left. Chief Marketing and Communications Officer Keith Weed said he understands why "headline risk" led some to do so. "I'm sure lots of people got under pressure from their boards and bosses, which were all genuine concerns," he said.

But Weed said YouTube already had adequate safeguards "some advertisers weren't using to the extent they could."

"I believe it's better to stand by a partner and work through a problem than walk away," he added. "Plus, you wonder how some of these brands can ever go back, because no one can give you a 100% guarantee of safety."

He also said the brand-safety controversy performed a service for marketers, by exposing that cheaper programmatic ad buys often come at the price of appearing on unsavory or irrelevant content.

A tougher knot
And, according to one marketer whose company audited a thousand YouTube URLS where its ads had run, brand safety isn't as big a concern as irrelevance.

"We found big relevance issues," the executive said. "A lot of inventory was going to the wrong place."

Alcohol marketers' ads often display in the wrong places, according to Andrew Serby, marketing director at Zefr, which analyzes and targets video placements on YouTube for advertisers. The company has been doing more audits for brands amid the year's focus on brand safety.

"They were showing up in front of 'Minecraft' videos, Peppa Pig and all this problematic content where the only possible consumer is a kid," Serby said. "We've looked at several campaigns where the target is a 25-year-old male or a 35-year-old woman, and no matter what, they're running against Peppa Pig videos, because the algorithm goes to where the eyeballs are, and younger people aren't skipping ads, and if there's a shared device and you're demo targeting, mom hands the kid an iPad and they see it."

Searching for answers
Big marketers do appear to have gotten more selective, well beyond YouTube. The number of sites running P&G ads from January through May fell 33% from the period a year earlier, according to analytics firm MediaRadar. Sites with Unilever ads declined 11%.

Unilever, J&J and other marketers are also using tools such as Grapeshot, which uses customized keywords to target ads by context and audience segments across several major programmatic ad-buying operations.

Brand-safety worries increased business, said Grapeshot CEO John Snyder. There's no blanket solution because brand safety means different things to different brands, he said. Auto brands may want to avoid advertising around news about safety defects and recalls, he said. Others may want to avoid negative coverage about celebrities who pitch their products.

But Grapeshot only has so much reach. It has only some access to Facebook and none at YouTube. The company piggybacks on audience-measurement tags and software from the analytics company Moat, software, but those aren't used within YouTube or most Facebook placements outside Instant Articles. Moat is part of YouTube's fledgling third-party verification system, now under initial audit by the Media Rating Council, but the Moat tag with Grapeshot isn't part of that system.

"It's just arduous and cumbersome to set up anything inside Google," Snyder said. "They let these things happen because they're a media company, and they want to sell as many ads as possible. Basically, the tools are there. They're just not using them."

Duopoly faces off
Google has its own customized keyword buying capabilities, of course, along with other systems designed to put ads exactly where marketers want them.

"Recently we expanded our advertising policies, increased our ability to enforce those policies and added controls to give brands more choice over where their ads appear," a Google spokesman said in a statement. "We're also applying our most advanced machine-learning research to help us more quickly identify and remove extremist and terrorism-related content. We'll continue to work closely with both advertisers and creators to get things right."

He added that the third-party audience verification system Google is preparing to roll out includes brand-safety reporting.

The controversy may be giving Facebook an edge over Google, or at least YouTube. Executives of both digital behemoths recently presented on brand safety to the board of the Association of National Advertisers, according to one marketing executive who attended. Facebook appeared to have better safeguards, the executive said.

Still, marketers willing to pay the price can substantially increase their control over YouTube placements. More big brands are piling into the already crowded Google Preferred program, which offers ads on popular, "premium" YouTube channels and lets marketers buy seemingly safe content categories like "beauty & fashion" or "foods & recipes," said Eric Goldman, VP-product at Zefr.

But Preferred inventory is scarcer and pricier, subject either to category exclusives or more clutter from competitors, and leaves out lots of high-quality content, Goldman said. Also, even Google Preferred included Pew Die Pie, until his channel was removed from the high-end platform in February over anti-Semitic jokes he made.

Who watches the watchmen?
Zefr charges brands to help target video content more precisely across a wider swath of Google and other video placements. But certainty is hard to buy at any price.

When P&G talked to Zefr about targeting video ads using Zefr's analysis, it asked whether yet another outside party could in turn verify Zefr's results, according to Goldman.

While Zefr's automated system is "human trained," the only way to be absolutely sure is by hiring thousands of people to watch millions of videos, he said said. Google and Facebook are doing that, he noted, but to weed out objectionable content, not determine relevance.

~ ~ ~
CORRECTION: An earlier version of this article said P&G had tested Zefr. The company now says it has not done work for P&G.