Haunted by keywords: 41% of Halloween Articles Wrongly Blocked, Reveals Mantis Data

Mantis, a leading brand safety, suitability and contextual advertising solutions provider for publishers and advertisers, has today released data highlighting brands that rely on outdated keyword blocklists will be blocking more than 41% of Halloween content and scary movie reviews this spooky season.

According to the findings gathered from Mantis’ network of publishers, words such as 'kill' 'blood' 'horror' 'scary' 'attack' 'death' 'terror' are amongst those on agency blocklists. While in theory, blocking ads from appearing next to articles containing these keywords avoids potentially unsafe brand environments, in reality it also cuts brands off from perfectly brand safe and suitable ad placements next to content that their target audience is actively seeking out.

For example, blocking the word “bloody” would inadvertently also block the London Dungeons Halloween “Bloody Mary” experience and all articles that have listed it as a must-see Halloween activity. Similarly, advertisers will fail to reach audiences looking to spook themselves by watching the latest Halloween blockbuster films.  With a quarter of UK shoppers considering Halloween their favourite holiday and spend projected to reach £776 million in 2024, marketers risk reducing their advertising reach during this tentpole event. As a knock-on impact, this over-blocking also severely reduces advertising income for publishers.

Mantis, which leverages advanced AI technology to understand the context in which these keywords are used, measured over 1100 articles to unveil these findings. In stark comparison to the standardised agency blocklists, only 12% of articles were categorised unsafe when using Mantis’ Natural Language Understanding (NLU) and Contextual Targeting categorisation.

Cath Waller, MD of Advertising at Immediate Media, has experience of the negative impacts that blanket keyword blocklists can have on Halloween content. Immediate Media publication goodFOOD has seen a number of its pages promoting novelty Halloween cakes blocked due to phrases such as “garish bakes that ooze fake blood”, or “For a spooky fake blood effect, use berry syrup”, and “Halloween slash cake”. 

“Our Halloween content is incredibly popular with readers at this time of year, with our publications working hard to create content that inspires and grabs readers of all ages,” comments Cath. “Our readers are always looking out for creative Halloween ideas, and that so many of these articles are blocked by advertisers is incredibly frustrating. By relying solely on out-dated tech, advertisers not only harm their own ability to reach a wide range of audiences, but also stop our readers from having the best possible on-page experience. It's time for more advertisers to step up and utilise AI-powered safety solutions that more accurately assess the suitability of web pages.

Fiona Salmon, Managing Director at Mantis also commented: “With thrill-seeking consumers on the lookout for content related to Halloween traditions, events, and deals, brands aren’t just sacrificing lucrative advertising opportunities, they’re directly harming publishers. The industry has been zombie-like in its reluctance to waver from keyword blocklists, despite their obvious negative impact. It’s time for advertisers to wake up and embrace brand safety solutions that prioritise suitability without impacting performance.

Methodology 

Mantis conducted the analysis of over 1,147 Halloween-themed articles from its publisher network on October 6th using BigQuery to compare standardised agency keyword blocklists with its own AI-powered Natural Language Understanding (NLU) and Contextual Targeting solutions. The study focused on keywords like 'kill' 'blood' 'horror' 'scary' 'attack' 'death' and 'terror', commonly flagged by blocklists. Articles were classified based on keyword detection and context: Green for safe content, Red for unsafe with high confidence, and Amber for potentially unsafe but below the blocking threshold.

About Mantis

Mantis is a leading brand safety and contextual advertising solutions provider for publishers and advertisers. Launched by Reach in 2019, the Mantis suite of solutions create an efficient, more transparent advertising ecosystem for publishers, brands, and audiences. Mantis’ suite of AI-powered tools – including Brand Safety, Contextual Targeting, Recommender, and Natural Language AI – provide market-leading solutions in today’s complex digital landscape. Built by a publisher to help publishers, agencies, and advertisers, Mantis is the most powerful and precise solution for creating a better advertising experience for all.