Three companies define commercial attention measurement in 2026. Adelaide, Lumen and DoubleVerify each solved a real problem, each earned serious distribution, and each is a credible choice inside its lane. They also share a boundary that the category rarely states plainly: all three measure attention on paid advertising. The question of what a human pays attention to outside a paid impression, in an article, a search result or an AI answer, sits outside all three by design.
This is not a criticism of any of them. It is a map. If you are buying media, one of these three is very likely the right answer. If you are trying to understand attention as a property of your brand rather than your ad buy, the map has an edge, and it is worth knowing where the edge is before you sign.
Adelaide: attention as a media quality signal for buying
Adelaide built AU, a media quality metric that scores an impression on its likelihood to earn attention and drive outcomes. Adweek has called it the attention economy's most widely recognised metric, and the distribution behind that claim is now substantial. Adelaide's AU was folded into Nielsen ONE through the Outcomes Marketplace in October 2025, combining attention scoring with Nielsen reach data. In June 2026 AU pre-bid segments became available inside Amazon DSP, and from mid July 2026 AU began feeding DV360 Custom Bidding as an impression-level signal.
The direction is unambiguous. Adelaide is becoming an input to programmatic buying, a number that helps a bidder decide what an impression is worth before it is served. That is a powerful position and a specific one. AU is predictive and omnichannel, built to value inventory at the moment of purchase. It is not built to tell you how much attention a human gave a paragraph on your website, because that is not what a bidder needs to know.
Lumen: eye tracking as the ground truth for ads
Lumen, founded in 2013, measures attention with eye tracking. Its dataset is real-world consented gaze data from more than fifty countries, and from that panel it builds machine learning models that predict how people look at ads across formats and devices without observing every campaign directly. Through 2026 Lumen pushed hard into connected TV: a Netflix partnership in March across five European markets, and an exclusive Teads deal in May for CTV HomeScreen measurement globally.
Lumen is the closest thing the industry has to a physical ground truth, because gaze is attention in its most literal form. The trade-off is structural and Lumen is honest about it. Eye tracking at true census scale on every user is neither possible nor consented, so the model is panel plus prediction: measure a representative sample precisely, then extrapolate. That is methodologically sound and it is the correct design for advertising, where the unit of analysis is the ad placement. It also means the measurement lives where the panel and the models are pointed, which is at ads.
DoubleVerify: attention as one input in an outcomes machine
DoubleVerify is no longer an attention company in any narrow sense, and that is the point. After acquiring Scibids in 2023 and Rockerbox in early 2025, DV became a full-funnel measurement and optimisation platform where attention is one signal among fraud filtration, viewability, suitability, attribution and AI-driven bidding. DV's own framing is a compounding loop: quality signals feed optimisation, optimisation feeds outcomes, outcomes feed back into quality.
For a large advertiser already inside the DV ecosystem, this is genuinely valuable, because attention stops being a standalone score to admire and becomes a lever wired into activation. The consequence is that DV's attention signal is defined by its role in that machine. It exists to help value and optimise media. It is not designed to answer what a human read, or which sources shaped a brand's reputation, because those questions do not feed a bid.
The boundary all three share
Adelaide predicts attention to value impressions. Lumen measures gaze on ads through a panel. DoubleVerify treats attention as one input to an outcomes engine. Three different methods, three different strengths, one shared frame: attention is measured on paid advertising, and the measurement is built to make media buying better.
That frame is correct for an enormous and important problem. Global brands waste vast budget on impressions no human attends to, and these three companies materially reduce that waste. Nothing in this article suggests otherwise.
But a brand's attention is not only spent inside its ad buy. It is spent in the editorial a journalist writes, in the video a creator posts, in the search result a user skims, and increasingly in the answer an AI model generates when someone asks it to recommend a product. None of that is a paid impression. None of it is visible to a tool built to score media quality for buyers. As search shifts toward zero-click answers and AI systems become the surface where brands are first described, the share of brand attention that lives outside paid advertising is growing, and the three dominant tools are, by design, pointed elsewhere.
Where a fourth approach sits
A different design starts from a different question: not what is this impression worth, but how much real attention did this content earn, wherever it lives. That requires measuring behaviour directly rather than predicting it from a panel, at census scale rather than through a sample, across owned pages, earned media and AI answers rather than only paid inventory.
This is the design behind BAX. It measures behavioural signals from real users at census scale, applies one mathematical model across web content, video and AI responses, and treats a paid impression as one context among many rather than the only one. Where Lumen extrapolates gaze from a panel, BAX observes behaviour from the full population that actually visited. Where Adelaide and DV are built to improve a media buy, BAX is built to describe a brand's attention across every surface where it competes, including the AI answers that no advertising-attention tool observes at all.
The claim is not that this is better than Adelaide, Lumen or DoubleVerify at what they do. For valuing impressions and optimising media, those three are the reference standard and a behavioural platform is not trying to replace them. The claim is narrower and more precise: the advertising-attention category has an edge, that edge is the boundary of the paid impression, and the attention a brand earns beyond that edge is measured by a different instrument.
How to choose
If your question is which impressions to buy and how to optimise a media budget, choose from the three. Adelaide if you want a predictive quality signal wired into programmatic bidding. Lumen if you want gaze-based ground truth, especially in video and CTV. DoubleVerify if you want attention folded into a full outcomes and attribution stack you already run.
If your question is how much attention your brand earns across everything you publish and everywhere AI describes you, none of the three is built for it, and you are looking for a different class of instrument. Know which question you are asking before you choose, because the tools are not substitutes. They measure attention on different sides of the same boundary.