Three years ago, mainstream news brands reported on marketers funding disturbing content, hate speech, terror propaganda, misinformation and vile filth on the open web – and the brand safety panic ensued. Today, huge lists of keywords assembled to bluntly sidestep such tricky topics are also demonetising important conversations about race, George Floyd and the Black Lives Matter movement for the very same media that highlighted the problem. The solution sure had some ugly side effects.
Keyword blocklists, sometimes thousands of words long, contain terms like ‘murder’, ‘bomb’ and ‘Isis’, and are designed to keep brands away from bad news, atrocities, murder and violent crime, so as not to be damaged by association. But terms like ‘LGBT’, ‘black’, and most recently ‘coronavirus’ have also found their way onto these blocklists, depriving publishers of a way to monetise their painstaking coverage of some of the biggest news stories in a generation.
Typically, brand safety tools cross-reference URLs for keywords on the blocklists. But without greater context, and with many marketers erring on the side of caution, the web is being shaped into an unsustainable place for numerous types of media.
Earlier this month, Vice told Ad Age that Black Lives Matters content saw 50% lower CPMs than the site average in the aftermath of the George Floyd murder. Vice can take the hit, but can black media focused on the topics of race, racism and the Black Lives Matter movement? Blunt force blocking of the coronavirus keyword has already cost tens of millions to the UK news industry alone. And LGBT titles have long suffered from brands tiptoeing around topics like “same sex” or “lesbian”.
Cheq estimates that publishers lost $2.8bn in ad revenue in 2019 due to keyword blocking.
What’s it all in service of?
Publishers are not paid for blocked inventory – and that includes premium titles like the New York Times and Wall Street Journal. Publishers that register high numbers of post-buy blocks can even be taken off campaigns entirely.
Chris Kenna, chief executive of Brand Advance and part of the Conscious Advertising Network, says: “It's all well and good posting a black square on your social media but if you're directing your spend away from the very people that are telling these stories it can seem like a hollow pledge.
"The dangers are very real – even outside of editorial media. I recently posted a livestream from a Black Lives Matter protest with my partner. People posted the N-word beneath it. A brand does not want to be next to comments that are racist. This can happen in both editorial and personal content. There are always going to be opposing views to editorial content.”
Brand Advance is on the hunt for brands that are keen to appear next to the positive waves of the Black Lives Matter movement. That'll be differentiated using deeper contextual analysis.
It’s not all doom and gloom. In recent years, Kenna notes that there has been a “dramatic” surge in ad spend against BAME, LGBT, 50+ and disability media as brands look to better address their consumers.
Dr Augustine Fou, the 'ad fraud investigator', dubs brand safety tech “blunt instruments” and says Black Lives Matter avoidance was no technical oversight. In fact, he is sceptical of the brand safety risk the news environment poses generally.
He warns that overzealous blocking sends spend downstream “to fraud sites, fake news sites, fake recipe sites, porn sites or piracy sites… where you're lucky if those ‘at scale’ audiences are even a little human.”
In response, publishers are rushing to solve the so-called brand safety problem. Reach, for example, has been building its keyword blocking tool Mantis with IBM Watson and the goal is to create an algorithm that relies less on guesswork and can read the actual page. It is motivated by the need to make its huge news ecosystem not just safe, but desirable to advertisers.
Last week, Mantis partnered with diversity-focused media network Brand Advance to address the ‘BLM’ blocking movement. Benjamin Pheloung, general manager of Mantis, blames “unsophisticated tools” for the issue.
“Almost everything about the Black Lives Matter movement is going to fall afoul of some kind of blocking. George Floyd's murder will be caught by segments blocking violence. Concerns about gatherings spreading coronavirus will be blocked by segments looking for Covid-19. Perhaps most perniciously, segments blocking hate speech will also be triggered when discussing racism."
Hard news articles “typically have a longer dwell time and deeper engagement than 'soft' news content,” says Pheloung. But for that same virtue, brand safety tools that scan content discriminate. Longer articles are more likely to have words that throw up false negatives.
“News publishers create the densest pages out there. Sites with lists, carousels of images and the like become the default home for open marketplace programmatic spend.”
At the bare minimum, marketers are going to have to review their blocklists and learn where the blockages occur. They must weigh up the necessity of the block against the social good of supporting journalism in tricky categories like Black Lives Matter and reaching consumers who are reading this material.
Nandini Jammi, co-founder of ad spend activists Sleeping Giants, and her business partner Claire Atkin, started Check My Ads, a consultancy to audit brands’ media buys across the web. Jammi told FastCo this week: “[Agencies] place the ads wherever because they get paid on impressions. I’m seeing the worst sites on the internet being monetised by these ads, and when I flag it to these companies, they remove it immediately.”
Are a few innocent but unfortunate clashes of ads with negative news stories worse than funding the grimmest corners of the internet?
At the start of April, IAB UK shared some guidelines for advertisers, urging them to reengage with the news environment and even prepare a simple PR line for any unfortunate juxtapositions that occur.
“We are advertising around this content on trusted news sites in order to support the content they are producing at this critical time,” read the disclaimer.
Marc Guldimann, founder and chief executive of adtech firm Adelaide, which measures the ‘quality’ of select media, says: “I don't think brands are doing this on purpose... and in fact, they are rarely in charge of their own setup."
He thinks that there is "very little content on premium publishers that can harm brands.”
Furthermore, the box-ticking exercises of the media buy have historically been disconnected from the impact of that spend.
Guldimann says: “It's very hard to attribute good things in advertising, but very easy to see bad things... so agencies are constantly trying to cover their asses – brand safety vendors play right into this fear. It's hard for someone to understand the impact of clicking a box in a DSP or adding a term to a blocklist. The distance created by technology makes it more likely that people will do socially unacceptable things, like block ‘black people’ as a keyword.
“Brand safety vendors exist because of fear.”
And to any publishers really taking a hit from these measures, he urges them to serve numeric URLs these tools can’t scan as a last resort.
What do the tools say?
It is not an invisible problem. Operators of these tools have for years issued advice to marketers.
Criteo says that “advertisers play an essential role in funding the free flow of information, especially the news. We encourage advertisers to consider this power and the impact of their actions when making tough decisions. In a time like this people need to be able to access accurate, reliable news for the benefit of the greater good.”
IAS also puts the burden on the marketer using the tool: “At the end of the day, the marketer controls what they block against. No one knows their brand better. Try not to treat a keyword blocklist as something to grow – instead, revise it regularly. Your keyword list should be the most up-to-date asset in your arsenal.”
Meanwhile, Double Verify claims that these measures, as flawed as they may be, “enable brands to continue to buy content in broader categories like politics and news.”
It says that “news avoidance” is a bigger issue and that less than 2% of impressions on the top 100 US news sites are blocked solely due to blocklists.
"Used in isolation, keyword blocking is a blunt technology that does not differentiate between such words being used in a negative way, in hate speech vs the context of a news article," says Tina Lakhani, head of ad tech at IAB UK.
"Context is key and contextual targeting and brand safety technology has come a long way – nowadays using methodologies to understand semantics with AI and natural language processing."
Marketers have been so obsessed with chasing their target audiences at the lowest possible CPCs and CPAs that less attention has been paid to where they are talking to them. "By doing so, we have taught our algorithms to value the lowest costing inventory, at the detriment to the value and relevance of the content and context we are appearing against," Lakhani says.
She urges brands to set up direct relationships with publishers that they want to support and work with. There, they can set hard rules on what they do, and don't, want to appear next to.
She concludes: "Ultimately, brands have a central role to play in actively seeking out and funding news that speaks to minority groups and diverse audiences – such as publications for black or South Asian people or the LGBT+ media.
"These publications can only thrive if advertisers are actively including them within their strategies, rather than just attempting not to exclude them."