The DOOH Audience Measurement Gap
DOOH audience measurement has historically relied on panel-based impression estimation: extrapolate from a sample of people who report being in a geographic area, apply venue multipliers, and deliver a modeled reach number. The IAB's 2024 DOOH measurement guidelines formalized this approach, but industry practitioners understand the core limitation: the person-count and attention signals are estimated, not observed.
The 731 sensing-enabled screens in the Trillboards network take a different approach. On-device computer vision — running on the local hardware, subject to privacy safeguards including immediate face de-identification before any data leaves the device — produces real-time audience attribute signals. These signals are classified using two taxonomies: IAB Audience Taxonomy 1.1 (segtax=4) for attributes that map to the standard, and the Trillboards proprietary taxonomy (segtax=600) for CV-unique attributes that have no IAB equivalent.
What segtax=600 Signals Are
The segtax=600 taxonomy covers six attribute classes that can only be produced by on-device computer vision — they cannot be inferred from location data, time-of-day modeling, or panel surveys:
Intent Stage Distribution
The intent stage signal is the most commercially significant segtax=600 attribute for direct-response DOOH buyers. Unlike location-proxy intent models (which infer stage from past GPS history), the CV-derived intent stage is a real-time behavioral signal — classifying current micro-behavior patterns (deliberate browsing posture, extended dwell, item inspection) against a trained model.
Across the 731 sensing-enabled screens with intent-stage coverage (71% of the sensing fleet), the distribution skews toward consideration (38% of observed audiences) and awareness (31%). Decision-stage audiences (12%) are concentrated in retail and QSR contexts where in-store purchase is the immediate available action. Post-purchase audiences (5%) appear predominantly in fitness and food service contexts — consistent with post-workout recovery or post-meal dwell behavior.
Attire Archetype — What It Tells Buyers
Attire archetype is not a proxy for demographics — it is a signal about context and purchasing context. A screen in a corporate lobby during business hours showing predominantly "business" attire (dress shirt, blazer, formal trousers) has a materially different contextual environment from the same physical location at 6:30 PM when attire shifts toward "casual" and the occupancy profile changes.
The distribution across the network is: casual (41%), business (29%), athleisure (18%), formal (6%), streetwear (4%), other (2%). Athleisure concentration — at 18% network-wide — reflects the fitness venue share of the network (17%, per Chapter 1) and is essentially a direct readthrough of venue mix.
IAB Audience Taxonomy 1.1 — 1,558 Mapped Nodes
The segtax=4 (IAB Audience Taxonomy 1.1) layer provides the DSP-readable audience classification. 1,558 distinct IAB AT 1.1 nodes have been observed active across the network during the observation window. This figure represents segments that received at least one impression — it is not the total segment catalog, which extends beyond 2,000 nodes.
The active segment density — 1,558 nodes from a 731-screen sensing subset — demonstrates that DOOH audiences are not homogeneous. A fitness screen in midday traffic and a quick-service restaurant screen in evening traffic can share a geographic market but have zero segment overlap. This heterogeneity is what makes real-time audience signals at the screen level valuable: the audience is not the venue type — it is the people in the venue at that moment.
What Changed Since 2025
The sensing-enabled screen count grew from an estimated 480 screens at year-end 2025 to 731 by May 2026 — a 52% increase in one year. Two changes drove this: firmware updates enabling on-device inference on existing hardware (without hardware replacement), and new venue onboarding prioritizing sensing-capable Android devices over legacy signage players.
The intent-stage signal matured significantly. Early deployments in 2025 had coverage on fewer than 50% of sensing screens; the 71% coverage rate in the current dataset reflects model improvements that extended the signal to lower-resolution camera configurations.
What to Expect in 2027
The sensing fleet is on a trajectory to exceed 2,000 screens by end of 2027 if current onboarding rates hold. More consequentially, the audio-with-diarization capability — detecting speech intent patterns without storing or transmitting audio — is cleared for production deployment. Speech-derived signals will add a sixth attribute class to the segtax=600 taxonomy, extending intent-stage confidence in noisy environments (transit, QSR) where visual signals alone are insufficient.
Implications for Advertisers
The practical value of segtax=600 signals is in creative optimization. A buyer who knows that 68% of the audience at a specific screen at a specific hour is in "consideration" intent stage with "casual" attire and "solo" cohort composition can deliver a direct-response creative. The same screen at a different hour showing "awareness" intent with "mixed_group" cohort composition is better served by a brand-awareness unit. Real-time audience signals make this optimization possible within programmatic buying — no direct deal, no manual creative trafficking.
Implications for Venue Owners
Venues with sensing infrastructure command premium CPMs. The measurability premium — documented across multiple IAB reports on DOOH measurement — reflects buyer willingness to pay a higher floor when impression quality is verifiable rather than modeled. Venues that activate sensing via firmware update (no hardware replacement required for supported devices) typically see CPM floor increases within 30–60 days as DSP algorithms adjust to the enriched signal.
External References
- IAB (2024). Digital Out-of-Home Measurement Guidelines v1.2. Interactive Advertising Bureau.
- IAB (2023). Audience Taxonomy 1.1. Interactive Advertising Bureau Tech Lab.
- eMarketer (2025). DOOH Audience Targeting: Beyond Location. Insider Intelligence.
- OAAA & Geopath (2025). Out of Home Measurement Transition Guide. Outdoor Advertising Association of America.