‘Sex’, ‘violence‘, ‘death‘, ‘alcohol‘, ‘slavery‘, ‘kill‘, ‘injury‘, ‘shoot‘, ‘disaster‘ and ‘bastards‘ – these are some of the keywords brands are choosing not to advertise against in 2020, enlisting automated tech to stop their ads from appearing around content containing these words.
With global digital ad budgets taking an increasingly big slice of the pie, brand safety is still high on chief marketers’ agendas. However, blacklists aren’t just protecting brands against misplacement: they’re also blocking active participles that feature in trusted publishers’ most important and most-read news stories, creating a deficit of support for hard-hitting journalism.
Top media owners are being penalised with blunt brand-safety tools to the tune of $3.2bn a year across the US, UK, Japan and Australia according to fresh research from real-time brand safety business Cheq. Its study found that the blacklists employed by publishers and buyers can include as many as 3,000 word and are blocking ads from articles, regardless of the context of the word.
Blocks on ‘kill’ and ‘murder’, for instance, have seen articles about Game of Thrones deaths demonetised. ‘injury’ and ‘shoot’ also feature on many lists, giving sportswriters a tough time. Mentions of race and sexuality are making life difficult for race-discussions and LGBT media, as told to The Drum last year. An exec at Hearst recently explained how stories about the Duchess of Sussex were being blocked by brands, because her title contains the word ‘sex’. In fact, this very article would be blocked on a site that employs overzealous blacklists after that sweary intro.
“In this current digital model, there is little attention to the quality of the content or the provenance of that content,” said Newsworks chief executive Tracey De Groose.
“There is no distinction between crafted journalism – that adheres to ethical standards, editorial codes, regulators and the law – and bedroom bloggers, amateur producers and at the extreme end, criminal content.”
De Groose argued that the impact of blocklists is much more significant for the newspaper industry, because journalists will always want to have the freedom to use the right words to tell their stories regardless of the consequences of ad revenue.
“The problem with this sudden knee-jerk reaction,” she explained “is that we have reduced a complicated issue like brand safety to the lowest common denominator.”
A race to brand safety?
It’s not just publishers losing out either, brands are limiting their audiences too.
Cheq said a single ad verification company blocked 9.8bn ad impressions last year. “Very high” and “significant” volumes of brand-safe inventory were curmudgeoned in the process. Some publishers estimated that they were losing 10% and 30% of ad revenue in a single campaign.
One ad verification provider which responded to the study said that in the last year alone it had seen a 250% increase in blocking. Another said blacklisted keywords were up 10%. Elsewhere, 43% of media buyers said they now explicitly avoided advertising next to the news.
Guthrie C Collin, chief analytics officer at Dow Jones, cautioned that this “race to brand safety” could see brands appear on less credible news sites with poor, or irrelevant, content.
De Groose argued that it was advertisers that created extensive blocklists, along with agencies, content verification companies and publishers.
Damon Reeve, chief executive of The Ozone Project, the publisher network boasting Reach, News UK, Guardian Media Group, Telegraph Media Group and Stylist, said the issue now lies in how media agencies are briefed on the issue.
“The responsibility to manage brand safety is generally delegated by a brand to an agency, who will operate on the conservative side to preserve the client relationship. They are overworked and so rarely will maintain blocklists which end up being long and cumbersome.”
Reeve said the current reliance upon keywords is disproportionately (and ironically) penalising, sites with a greater journalistic impact who run high quality, editorially-governed environments.
“The knock-on effect is that advertiser budgets are pushed onto long-tail unregulated sites, perversely increasing the risk for the brand.”
Those managing the gates need a better understanding of the ecosystem, he added, saying that “natural language processing and contextual understanding” tech could also be used to apply a layer of nuance to the process and free up more inventory without costly or manual human intervention.
Bringing contextual tools up to scratch
With Google moving away from the third-party cookie in the next two years, the industry will put more emphasis on contextual targeting. An announcement that Reeve said will see an even greater focus on what contextual understanding and targeting can deliver for advertisers.
Ben Walmsley, commercial director of publishing, News UK blasts the “crude approach” currently employed, and said solutions need to be more sophisticated.
He explained how the process can be improved. “In a test scenario with the 2017 Manchester bombings, our adtech partner easily categorised the negative sentiment associated with an article reporting on the event.
“However, it recognised that an uplifting story about a fundraising concert for those affected had a positive sentiment and would have allowed the ads to be placed, despite the same negative words being present in the article.
“The most basic tools in use would block articles about Manchester United, recognising ‘Manchester’ as a negative word years after the event.
“Coupled with editorial discretion, technology working with context and semantics rather than simple keywords can achieve a quite accurate understanding of sentiment and emotion.”
However, the wheels fall off this if buyers add additional restrictions into the mix.
“There is a distinct advantage to be gained by those willing to invest time in customising their approach.”
David Hayter, head of digital at Stylist, has seen feminism and LGBT on blocklists – these are fairly common content pillars at the title.
“Often these articles are celebratory pieces that brands should want to advertise around. Instead, by using the current technologies on offer what they are saying is that we don’t value certain audience groups.”
The contextual tools are not up to scratch according to him. Stylist’s solution also factors in readers’ responses to the editorial. This seems like an easy enough way to analyse sentiment, see what the humans are doing.
“If we write an article about a tech brand and that article has both a negative sentiment and the overriding emotion is anger, then it would be understandable for that brand to advertise around the content.
“However, if we are writing about that brand and the overriding sentiment is positive and the main emotion is joy, then why wouldn’t an advertiser want to be against that content.” He noted that Reach’s Mantis tool, and Permutive, used by BuzzFeed employ similar measures.
“If the money does flow towards contextual targeting, in its current state I think this is a bad thing for the open internet. We’ll start to see very vanilla content across the web as publishers try to stay on the right side of the targeting tools.”
To the contrary, Nick Morley, EMEA managing director of Integral Ad Science said the blocking has been “over-zealous” and that advertisers must focus more on contextual targeting. By combining brand safety protection with customisation, advertisers can avoid risky content and proactively target suitable content, while also supporting high quality journalism.”
Justin Taylor, managing director UK of video ad firm Teads pointed to research indicating that it is the ad industry, rather than readers, that is worried about brand safety.
“Readers do not worry about this as much as we do as an industry. Where brand safety and protection is a must, brand suitability is a much more subtle and bespoke art.
“Publishers and platforms need to work with brands directly to understand their content tolerances and their audience needs. As a minimum, their brand safety and suitability guidelines setup should be dynamic, inclusive and reviewed before each campaign.”
