Programmatic advertising, with its cheap and easy targeting of individuals based on their online browsing habits, has appealed to brands since its onset. Yet the risks, while always known, have never really come to the fore, until now. Advertisers have been up in arms over the realization that their ads were showing up alongside racist and extremist videos on YouTube and other sites, and essentially underwriting hate.
While automation has its upsides, this is a time for human intervention. That means brands making more efforts to find safe spaces curated by humans, and YouTube using more human engineering to help match brands to better content. Not to mention users making judgments on what it means when big brands support objectionable content (even if they don’t realize it).
An Unfolding Disaster
A ticktock of events, including a damning Times of London investigation and decisions by the Guardian, AT&T and Verizon to pull their advertising from YouTube’s “hate” videos, shows just how big this problem has become for Google and the industry. The advertiser boycotts have continued with Johnson & Johnson, PepsiCo, McDonalds and even Havas Media Group following suit.
Adweek commissioned the market research company Survata to survey daily YouTube video viewers in a weekend online poll about whether they deemed ads as endorsements of the content; the results were both worrisome and relieving, depending on how you viewed them. Thirty-six percent of the 502 respondents said yes, they do interpret ads as endorsements. But about half of respondents also said that this didn’t make them change their opinions about these brands.
Regardless of what audiences may actually think, investment firm Nomura Instinet estimated the advertiser boycott could have damaging effects on parent company Google, and cost the search giant as much as $750 million in revenue. Google and other firms, meanwhile, have countered back. Google is tip-toeing around whether the problem is as big as everyone says and feels, and RBC Capital Markets said in a research update that the part of Google’s business affected by the YouTube fiasco is technically only about 10 percent of the company’s revenues. And the $750 million is also less than one percent of Google’s projected sales for this year, which means Google might not take a huge hit if it reacts fast enough.
Meanwhile, Google has altered its rules for what videos can carry advertising, and invested research efforts into improved AI to better understand the context and sentiment of YouTube videos. Increased scrutiny on its public performance likely means it’ll invest in this even more.
The JPMorgan Chase Experiment
With Google being one of the world’s biggest platforms, it’s difficult for most brands to simply opt out and boycott completely. But larger firms are experimenting with ways to bypass total automation and use human judgment to find the coveted safe brand spaces.
Take JPMorgan Chase. The company told the New York Times that it has started whitelisting websites that run its ads via Google. Prompted by the realization that an ad for Chase’s private client services had appeared on a site called “Hillary 4 Prison,” it ran a 30-day test period in which an intern manually clicked through 12,000 websites where JPMorgan ads were appearing, to see whether they were being featured alongside objectionable content. The intern flagged about 7,000 ads that were appearing where the company would prefer them not to be.
Now, JPMorgan has arranged a “human-checked” list of 1,000 YouTube channels it has deemed safe for its ads to appear. And it has dropped the number of websites showing its ads from 400,000 to 5,000. So what did it learn from the test? Performance appeared to the be same, even though the ads were displayed on far fewer sites. Chase’s CMO Kristin Lemkau told the Times that whitelisting was becoming de rigeur now at the bank, and that “at some point, a human is going to take a look.”
New Skepticism About Automation
JPMorgan’s move into whitelisting, the YouTube controversy and advertiser boycotts are stirring larger conversations on the necessary evolution of the digital advertising industry. It’s not just that the technology is failing or not yet up to the standard of human editors, but that relying on programmatic advertising allows networks to “grade their own homework.” And this kind of blind advertising is bad for business.
Outsourcing monitoring to third-party auditors and in-house ad networks, which I previously wrote about, is a potential solution. But the issue is that programmatic has become a growing default that’s hard to slow down. Forrester Research said in a report that only 29 percent of marketers in 2016 took steps to whitelist websites for their ads. And some smaller brands simply can’t manage the resources to do that.
Hany Farid, a professor at Dartmouth and senior adviser to the Counter Extremism Project, has also said that Google, Facebook and other major companies ought to deploy more human editors to filter content — because the hype around AI is just that: hype.
But the bottom line is that human intervention is costly, and would be a paradigm shift for the way the advertising industry has evolved in recent years. It may be the way to go, but it’ll take some time for it to become just as automatic as programmatic.