Lesson 5 of 5 · 7 min read · intermediate
Attention and the new metrics
Viewability says an ad could be seen; attention metrics estimate whether it was. Learn how attention is measured, new CTV currencies, and how to judge them.
Viewability was a big step forward, but its bar is low: half an ad on screen for a second. Two ads can both be "viewable" while one is ignored and the other is watched with the sound on. Attention metrics try to measure the difference, estimating how much human attention an ad actually received.
Viewability is like checking that a leaflet was put through the letterbox. Attention is like knowing whether someone picked it up, how long they looked at it, and whether they read it or dropped it straight in the bin.
How attention is measured
In 2025 the IAB and MRC published attention measurement guidelines, with a final version 1.0 released in November 2025, laying groundwork for future accreditation. They group methods into four families:
| Approach | How it works | Trade-offs |
|---|---|---|
| Data signals | Time in view, share of screen, scroll speed, audio on, interaction, from tags or SDKs on every impression | Scalable, but a proxy for attention |
| Visual tracking | Eye-tracking via webcams or phone cameras with consenting panellists | Direct, but small panels that must be modelled |
| Physiological and neurological | Heart rate, skin response, EEG in lab settings | Rich, but expensive and small |
| Panels and surveys | Asking people what they noticed and recalled | Meaningful, but slow and memory-based |
Vendors typically train a model on panel data (where eyes really went) and apply it to signals from every impression, producing a score per placement or a metric like attentive seconds per thousand impressions. Specialists and verification firms in the US, UK, Europe and Australia sell such scores, and buyers increasingly use them to choose inventory.
Why does it matter commercially? Because two placements with the same viewability rate and CPM can deliver very different attention. If placement A earns an average of 1 attentive second per impression at a $4 CPM and placement B earns 4 seconds at a $6 CPM, B is far cheaper per attentive second. Buyers who optimise this way often end up paying more per impression but less per real moment of attention, which tends to reward quality publishers with fewer, better ads.
Other metrics that matter now
- Video completion rate: the share of video ads played to the end. Useful, but on non-skippable and CTV formats it is often near 100% and says little.
- Reach and frequency: how many distinct people or households saw a campaign and how often, the backbone of TV planning and increasingly of CTV.
- Gross rating points: reach multiplied by frequency as a percentage of a population, the traditional TV currency.
- Outcome metrics: store visits, sales lift and brand lift, which connect exposure to business results.
New currencies for streaming TV
As viewing shifts to CTV, the old TV ratings model is being challenged. In the US, alternative measurement companies combine smart-TV data from automatic content recognition, set-top boxes and panels to measure audiences across linear and streaming. Many large buyers now transact on more than one currency. In India, Europe and elsewhere, broadcasters and industry bodies are similarly revising panel-based systems to include streaming. Different currencies can report different numbers for the same programme, so agree in advance which one counts.
How to judge a new metric
- Is it valid?Does the metric predict something you care about, such as recall, sales lift or brand lift, in independent tests?
- Is it transparent?Can the vendor explain the method, panel size and model, and is it following the IAB/MRC guidelines or pursuing accreditation?
- Is it comparable?Does it work across formats and platforms, including walled gardens, or only where the vendor has tags?
- Is it clean?Is invalid traffic removed before the metric is calculated?
The direction of travel is clear: from counting ads served, to ads that could be seen, to ads that were noticed, to ads that changed behaviour. Each step is harder to measure, and each step is more valuable, as long as the humans being measured are real.
Key takeaways
- Attention metrics estimate whether people actually noticed an ad, going beyond viewability's opportunity to see.
- The IAB and MRC published attention measurement guidelines in 2025, covering signals, visual tracking, physiological and panel methods.
- CTV is driving new audience currencies built from smart-TV, set-top box and panel data.
- Attention and completion can be faked, so always filter invalid traffic first and judge by outcomes.
Questions people ask
What are attention metrics in advertising?
Attention metrics estimate how much genuine human attention an ad received, beyond whether it was technically viewable. They use signals like time on screen, share of screen, audio and interaction, often calibrated with eye-tracking panels, to score placements or impressions. The IAB and MRC published guidelines for attention measurement in 2025. Buyers use them to choose inventory, alongside outcome measures like sales lift.
Is attention better than viewability?
Attention builds on viewability rather than replacing it. Viewability confirms an ad had a chance to be seen; attention estimates whether it was actually noticed and for how long. Vendors and some independent studies report that attention predicts outcomes better, but it is less standardised and harder to audit. Use viewability as a minimum, attention to choose better placements, and invalid traffic filtering underneath both.