The advertising industry’s relationship with contextual targeting has followed an interesting arc. In the early days of online advertising, contextual was the default: you placed an ad on a cooking website because you were selling cookware, not because you had identified individual users who had recently searched for frying pans. The rise of programmatic advertising and third-party cookies changed that, making user-level behavioural targeting the dominant paradigm and relegating contextual to a fallback for inventory that couldn’t be targeted behaviourally.

Now, with the third-party cookie largely gone and privacy regulation significantly constraining what behavioural targeting can legally do, the industry is rediscovering what contextual advertising does well - and in many cases, discovering that it does it better than the behavioural alternative.

Why contextual fell out of fashion

The appeal of behavioural targeting, when it emerged, was straightforward. If you could identify that a specific user had been researching new cars in the past week, you could show them a car ad wherever they went on the internet, not just on automotive sites. The advertiser was paying for the audience, not the context, and in a world where most browsing happened on broadly general-interest sites - news, social media, entertainment - context was often a poor proxy for intent.

Contextual targeting also suffered from crude implementation. Early contextual systems matched ads to pages based on keyword matching, which generated notorious mismatches: violence-related news stories serving alcohol ads, articles about cancer serving pharmaceutical ads targeting healthy people. The technology worked in principle but was poorly executed in practice.

What changed

Two things happened in parallel that have substantially rehabilitated contextual advertising. First, the technical quality of contextual analysis improved dramatically. Modern contextual targeting systems use natural language processing and semantic analysis to understand the meaning, tone, and context of content far more accurately than keyword matching could. They can distinguish between an article about overcoming addiction (probably not the right environment for alcohol advertising) and an article about wine regions in Burgundy (obviously relevant to drinks brands).

Second, the data on behavioural targeting quality has become much more scrutinised. The programmatic ecosystem’s self-reported performance metrics were always vulnerable to manipulation, and as independent measurement has improved, the genuine incremental value of behavioural over contextual targeting has turned out to be much smaller than the industry claimed. A study by researchers at Vanderbilt University, published in 2019, found that behavioural targeted ads commanded a price premium of around 22% over non-targeted ads - but that the actual performance improvement they delivered was considerably smaller, and in many cases negligible.

More recently, research conducted in the post-cookie environment has consistently found that contextual advertising performs comparably to or better than cookie-based behavioural targeting when the contextual relevance is strong - which, for niche publishers with well-defined content verticals, it consistently is.

The niche publisher advantage

For niche publishers, the rehabilitation of contextual targeting is genuinely good news. A site focused on a specific subject area - independent financial advice, specialist gardening, professional audio engineering, historic architecture - has a content environment that is extremely well-defined from a contextual advertising perspective. Every page provides clear signals about the audience’s interests and intent.

This is exactly the environment in which contextual advertising performs best. An advertiser selling specialist gardening equipment doesn’t need to track individual users across the web to find their audience - they need to be on specialist gardening sites, which their audience visits specifically because they are interested in gardening. The targeting is done by the editorial positioning of the site, not by surveillance infrastructure.

The CPMs available for contextually targeted inventory in well-defined niche verticals are, in our experience, considerably higher than the generic programmatic fill that most niche publishers have been receiving from standard ad networks. When we onboard a new publisher partner and move their inventory to direct, contextually relevant advertising relationships, the revenue improvement typically comes from two sources: higher CPMs per impression and better fill rates, because advertisers actively want to be in front of an engaged, relevant audience.

Combining contextual and first-party data

The most sophisticated approach for niche publishers in 2025 is not purely contextual but rather a combination of strong contextual signals with first-party data about reader behaviour and preferences. For a publisher with a registered reader base, the combination of “this page is about [topic]” with “this reader has been engaging with [topic] content consistently for six months” creates a targeting signal that is simultaneously very accurate, genuinely consented (because the reader chose to register and has agreed to their reading behaviour being used for personalisation), and fully privacy-compliant.

This kind of first-party data advertising is available to any publisher who invests in building a reader relationship - a login or newsletter subscription - and configures it correctly from a consent and data protection perspective. It doesn’t require third-party cookies. It doesn’t require data management platforms that aggregate surveillance data. It requires an engaged audience, an honest consent process, and an advertising partner who understands and values it.

The irony is that this is a model that gives niche publishers a genuine advantage over general-interest ones. A large news site with millions of daily visitors has a harder time developing the kind of specific, deep reader relationships that a specialist publication can build with its much smaller but more loyal audience. In a contextual and first-party data world, depth of engagement beats breadth of reach.