Earlier this month, advertising analytics company Adalytics released new research which gave a transparent view on brand safety segments, and keyword blocking tactics used by big brands, facilitated by the big tech partners in the space – namely, Integral Ad Science and Oracle’s Grapeshot. The research called out where brand names were being used in the naming convention for brand safety segments, and showed how these segments link to the list of keywords that informs them, giving the industry much needed transparency into the blunt and outdated keyword blocking strategies still predominantly used by brands and agencies.
There’s nothing new or shocking about these keyword blocking strategies, and 2020 in particular showed publishers that applying brand safety using a keyword blocking strategy alone can result in huge revenue loss. Back in early 2020, Mantis launched to focus on solving these issues for publishers, and move the industry towards a more intelligent, AI-led solution across brand safety and contextual targeting.
The Adalytics research has shed light on a number of issues with keyword blocking. From award winning and notable journalists having up to 97% of their content blocked, to publishers like Vice.com having over 50% of content blocked by a single vendor, it’s clear that as an industry, we need to move past these outdated methods to ensure we are providing a fair and safe open web for writers, readers and buyers.
What is certainly the most concerning aspect of this report, is the glaringly open way in which brands are blocking keywords which relate to religion, race, sexuality, gender identity and ability. The evidence is readily available across most brand blocklists, and now in the airtable provided by Adalytics in which the following words can be found: Muslim, Hindu, Sikh, Jewish, Christian, Gay, Lesbian, Bisexual, Transgender, Queer, Disabled, Disability, Pregnant, Period, Woman, Girl, Feminist, Asian, Chinese, Pakistani….
So these heavy brand blocklists are not only causing loss to publisher revenue through their sheer size, they can also be hugely prejudiced. If we take the word ‘Transgender’ as an example, blocking this word doesn’t just cut off safe and monetisable content, it cuts off an entire community who’s stories and voices need to be heard, not only in the LGBTQIA+ Publishing community, but in mainstream media and national news as well.The same can certainly be said across all minority communities.
In every conversation had about this subject, we are all rightfully angry and concerned about discriminatory keyword blocking. It is not new and has been called out across the industry numerous times. The fix is simple, right? Brands should review blocklists and remove the words and phrases which are so obviously discriminating and cutting off advertising to entire communities and diverse audiences (audiences which are not only key, but a true representation of global users).
So why is it that today, we will still find these words on almost every keyword blocklist provided by a brand? Some say it’s laziness, some say agencies are hesitant to approach their brand clients. An IAB study talking to agency planner buyers around their satisfaction in brand safety showed that only 7% were satisfied with their current brand safety strategy, but 83% admitted they rarely challenge their clients around techniques and tools used. So it seems everyone is aware of the problem but not many are willing to act.
This is not the case for some strong voices across the industry, who are willing to publicly call out where discriminatory blocking lives and how it’s affecting quality journalism. In August 2020, Marketer advocate and co-founder of Check My Ads Nandini Jammi challenged Integral Ad Science on it’s ratings for lesbian sex education content, and Black Lives Matter content, proving that these tools are not yet sophisticated enough to identify content written about/ for minority communities without labeling them as unsafe or negative. If brand safety strategies continue to be based on a keyword blocklist, then we’re going to keep seeing prejudiced results.
We know that a lot of these blocklists have been built for vast open marketplaces where the risk of running on unsavory content (such as pornography or extremist content) is high, but we have to change the perspectives of brands to understand that words like ‘lesbian’ don’t equal porn, and ‘islam’ does not equal terrorism – and that there are intelligent tools out there (like Mantis) that can identify truly unsafe content based on context, rather than keywords.
Despite the efforts of those like Jammi, there is still an air of mystery as to why something that seems so simple to change is not getting the focus it deserves. That is why at Mantis, we hold a strong position against discriminatory keyword blocking. When requests are made to implement a keyword blocklist, we perform a human review to call out and remove any discriminatory keywords, and communicate this back to the agency – in the hope that the message will be passed on to the brand (this is where we have experienced a blocker up the chain and have received agency feedback to say the message was passed along, and is often met with no change). Despite the pushback, we will not allow keywords or phrases relating to religion, race, sexuality, gender identity and ability to be the reason for blocked content. With Mantis being an AI-led technology, brands are able to implement blocking at a level which provides higher accuracy, without cutting off important content.
In September 2020, Mantis joined premium mobile ad format company Picnic’s ‘Beyond Social’ event to discuss how we are fighting prejudicial exclusion lists with AI technology, and how to use intelligent brand safety tools for a diverse open web. During this session, we shared a number of ways in which the publishing industry can challenge discriminatory blocking.
- Challenge Brands: Here you can apply a ‘see it, say it, sort it’ approach. There have been many cases where a brand is simply not aware of what is on the blocklist, due to the lengthy nature of the keywords themselves, and the fact that the blocklist was probably started back in the early 2000’s.
- Own your Brand Safety Strategy: Chose a partner that has strong values against prejudicial blocklisting, and strongly consider a publisher first solution
- Brands to work closer with publishers: Brands can benefit from gaining a broader understanding of brand safety and that a ‘one size fits all’ strategy doesn’t work for most publishers. For example, blocks on Enterntainment sections are going to massively differ from those required on breaking news content (i.e a murder on Eastenders is a very different sentiment to a report on a real life murder.)
If you would like to hear more about how Mantis is fighting against discriminatory blocking, please reach out to Emily Britton, Head of Publisher Development for Mantis at emily.britton@reachplc.com.