Attention Metrics

Attention metrics are a second-generation measurement layer that sits on top of viewability. Where viewability asks "was the screen on and visible?", attention asks "was a person actually looking at it?" The distinction matters because a screen in a busy gym lobby may register 100% viewability but 30% attention, most people are looking at their phones.

The IAB's 2025 attention working group converged on three measurement primitives: dwell time per person, gaze direction during dwell, and head pose during dwell. Combined, these yield a per-playout "attention score" that's roughly correlated with brand-recall lift studies. The DPAA's parallel framework adds emotional valence (smile/neutral/frown) as a fourth primitive.

Implementation is on-device or it doesn't scale. Trillboards screens run a sensing SDK that does face detection, head-pose regression, and gaze estimation locally, no faces ever leave the device, only the aggregate counts and scores. This is also the GDPR-clean implementation: aggregate counts and discrete attention buckets aren't personal data, while raw face vectors would be.

Buyer demand for attention data is rising fast, eMarketer reported 41% of CTV/DOOH spend in 2026 used attention as a primary KPI, up from 12% in 2023. Expect attention to displace viewability as the headline measurement metric by 2028.