Virtually all advertisers are acutely aware of brand safety issues these days – from their ads showing up next to terrorist videos on YouTube or ads showing up on fake news, hate speech, or disinformation sites. So, they’ve been increasingly paying for ‘brand safety’ technologies that are supposed to help keep their ads away from these situations. But do those technologies do what they claim to do? No, and it’s even worse than that. 

Brand safety detection tech doesn’t work well

Previous examples have shown that brand safety tech was responsible for blocking ads on the homepages of major newspapers like the New York Times and Wall Street Journal, due to the presence of keywords like ‘COVID-19’ or ‘coronavirus.’ Despite their claims of advanced machine learning and ‘semantic’ understanding (what words mean), the evidence suggests that ads were blocked based on rudimentary keyword lists. Furthermore, ‘content classification’ of entire websites is prone to error (see examples below); even classification of individual webpages for brand safety may be entirely wrong. The word ‘blood’ in an article on a medical journal should be marked as ‘safe’ while the word ‘blood’ in other contexts may be marked as ‘unsafe.’ 

Source: LinkedIn


Then, there’s the issue of the actions taken as a result of the safety classification. For example, keyword blacklisting is a common practice where ads are blocked from a page if it contains a keyword on a blacklist. ‘Shooting’ is one of the most common blacklisted terms. However, it could mean ‘shooting stars’, ‘shooting hoops’, or ‘shooting a photo’ and ads should not be blocked. Recent research by Adalytics confirms how problematic and inconsistent brand safety detection is. For example, an analysis of nearly 26,000 articles on a mainstream news site – wsj.com – revealed that four in ten articles were classified in completely opposite ways – safe/unsafe versus unsafe/safe – by two different brand safety vendors.

This is similar to other adtech data issues like classifying the same user as both male and female: it’s just bad data. Finally, brand safety tech is severely limited when it comes to text inside images. Even though some vendors claim to be able to detect unsafe words inside images using optical character recognition, most advertisers choose not to incur the extra expense; so, in effect, the brand safety tech is not protecting advertisers against brand safety issues on sites with predominantly image and video content. 

Defunding real news

In addition to not protecting advertisers from unsafe situations, there’s even more harm coming from the use of brand safety tech: the systematic defunding of real news. That means that ads are blocked on major, mainstream news sites because of keywords like ‘Black Lives Matter’, ‘protests’, ‘coronavirus’ etc. When ads are blocked on the homepages of these sites, the highest trafficked page of the site, it causes significant negative impact on the ad revenue of these news sites. But the ad dollars have to be spent, so those dollars and ads flow to less savory and higher risk sites like fake news, hate speech and disinformation sites. Adalytics data shows that between 10% and 50% of articles on mainstream news sites are marked as ‘unsafe’ by brand safety tech vendors, resulting in significant harm to those news publishers, i.e. revenue loss.  

But advertising on ‘the news’ is not bad. In fact, it’s better. Research shows the benefit of advertising on news sites. “Consumers feel that ads in a hard news story are more trusted. Average ad dwell time is 1.4 times higher in a hard news environment (45 seconds versus 32 seconds),” explains Michael Follett, managing director of Lumen Research, adding that “Ads on news sites get more attention than the same ads on other sites. This is, we believe, because people spend a long time engaging with the content on newspapers and so attention spills over from the content to the ad.”

Funding fake news, disinformation, piracy

If the ads are blocked on real news sites, the ads and dollars flow to fake news sites. A special report from NewsGuard, published January 14, 2021, showed that more than 1,600 major advertisers – from P&G and Disney to The American Cancer Society and Harvard University – ran programmatic ads on ‘untrustworthy websites’ spreading misinformation, inflammatory content, falsehoods about Covid-19 etc. The ad dollars from mainstream advertisers flowed to fake news and disinformation sites, providing them with funds to survive and expand further. This is leading to real-world harm, like the events of the past two weeks. The Global Disinformation Index has been documenting dozens of examples of major advertisers ads and ad dollars flowing to disinformation sites, via ad tech platforms. Furthermore, fake sites disguised as news sites are also diverting ad revenue to themselves.

See: These Hugely Popular Local News Sites In The US And Canada Are Fake. And a study by White Bullet showed that ad supported piracy sites are bigger and more profitable today than ever before, all enabled by programmatic ad tech. 

The utter lack of efficacy of brand safety tech is self-evident. Despite advertisers paying millions for these services, ‘all of the above’ is still happening, and it’s worse than ever. 

Recommendations for marketers

What can marketers do once they realize that brand safety technologies are not protecting them as they claim and are, in effect, enabling more ad dollars to flow to fake news and disinformation websites?

  • An obvious first step is to reduce spending through programmatic channels, which send their ads to hundreds of thousands of websites, including fake news, hate speech, porn, and disinformation websites.
  • Marketers should increase spending through ‘direct buys’ with mainstream publishers, including major news sites. Marketers can do so by specifically buying direct deal IDs set up by specific publishers in specific exchanges. This way, marketers also avoid the problem of domain spoofing where fake sites pretend to be mainstream sites. If marketers do this, then they wouldn’t need to pay extra for brand safety protection tech.

There is hope on the horizon. Due to activism by shareholders and groups like Open MIC (Open Media and Information Companies Initiative), large advertisers like Home Depot and their agency Omnicom are starting to investigate if their ads funded misinformation (hint, they did). 

Finally, common sense will tell you real human consumers visit news sites, they pay more attention when on such sites and they don’t think negatively about the advertisers even if the ads run next to news stories about sensitive topics. Consumers can recognize the difference between the ad and the news. So marketers should ask themselves whether it’s worthwhile to buy low-cost ads through programmatic channels, only to have to pay more for brand safety tech, but still inadvertently fund fake news and disinformation.