What is brand safety in advertising?
Controlling what content an advertisement appears alongside, so that a brand is not associated with material that damages it — a concern created by programmatic advertising, where the advertiser does not choose the placement.
The problem. In programmatic buying, advertisements are placed automatically across enormous numbers of sites and videos in real time. The advertiser buys an audience, not a placement, and has no direct knowledge of where the advertisement will appear. It can therefore end up beside content the brand would never sponsor.
High-profile incidents in which major advertisers found their advertising beside extremist and harmful content prompted large-scale withdrawals and drove the whole apparatus that now exists.
The tools used:
Blocklists of sites and keywords to avoid.
Allowlists of approved sites, which is far safer and much smaller in reach.
Content classification by third-party verification providers, categorising pages and videos.
Contextual analysis, assessing meaning rather than matching words.
Industry frameworks defining categories of unsuitable content and levels of tolerance.
The serious unintended consequence: news defunding. Crude keyword blocking penalises journalism, because news reports contain the words describing the events they report. Blocklists containing terms such as violence, death, disease or conflict systematically demonetise legitimate reporting — and studies have found substantial proportions of news inventory blocked this way.
The effect is perverse: advertising revenue is withheld from accurate reporting about difficult subjects, while content farms writing about nothing in particular remain fully monetisable. Brand safety, applied crudely, funds trivia and starves journalism.
Brand suitability is the more considered successor concept — recognising that appropriateness is brand-specific and context-dependent rather than universal, and that a blanket keyword block is a poor instrument.
The other criticisms: blocklists are opaque and rarely reviewed; they can disproportionately affect content from and about marginalised groups; and they reduce reach without evidence of commensurate benefit.
Where the genuine risk lies is now widely argued to be less about adjacency — which research suggests consumers understand is automated — than about funding harmful content through advertising revenue.