Attention metrics

Attention metrics estimate how much human attention an ad received, from signals such as time and share of screen, interaction, eye tracking or panels, going beyond the yes-or-no test of viewability.

Viewability says an ad could be seen. Attention metrics try to say whether it was seen, and for how long. Vendors model attention from device signals, such as how long and how large the ad was on screen, scrolling, and whether sound was on. They calibrate those models with eye-tracking panels and sell scores for planning and bidding.

Until 2025 every vendor defined attention its own way. The IAB and the MRC released Attention Measurement Guidelines for public comment on 12 May 2025. A task force of more than 200 members drafted them, and comments were due by 12 July 2025 (IAB, 12 May 2025).

The MRC announced the final guidelines on 12 November 2025. They group measurement into four methods: data signals, visual tracking, physiological and neurological observation, and panel or survey methods (PPC Land, 25 November 2025). The guidelines are the basis for MRC accreditation audits of attention vendors.

Attention works best next to outcome measures. A high score does not prove a sale. It helps explain why two placements with the same viewability perform differently.

Example: Two video placements both show 70% viewability. A vendor's model credits one with twice the seconds of active viewing per impression. The buyer moves budget toward it and runs a holdout test to see whether sales follow.

Related terms: viewability, brand safety, incrementality, MMM, vCPM.

Authoritative source: IAB and MRC, Attention Measurement Guidelines (public comment draft 12 May 2025; final November 2025), listed on the MRC standards page.

Read the full guide: Viewability, attention metrics and brand safety.