PLATFORM
Canopy Platform Edge Devices AI Agents Privacy by design
INDUSTRIES
Industries overview Hotels & Hospitality Retail Smart Cities Construction Out-of-Home
PROOF
Case Studies Insights
TECHNOLOGY
Technology overview Edge Compute Data Sovereignty OAIC Compliance
COMPANY
About Us Contact
Sign in Get in touch
CASE STUDY

Anonymous audience measurement for a national OOH network

How we helped an ad network prove audience value without tracking individuals.

Case Study · Out-of-Home — National digital signage network

A national OOH network was selling digital screens on modelled traffic counts. Advertisers wanted a defensible audience number. The network wanted to give them one without tracking a single person.

Australian city street with a digital billboard

The problem

OOH measurement has been stuck for a decade. The alternatives — modelled traffic counts, mobile location data, panel surveys — each have their own credibility problem. Mobile location data, in particular, is legally uncomfortable in Australia post-2022.

Our deployment

We mounted a small camera above each digital screen, running a Canopy CU-1 in the equipment cabinet. The edge ran an audience agent: it counted people who looked at the screen for more than two seconds, estimated their age/gender bracket, and aggregated the numbers by hour. No individual was identified. No data left Australia.

The result

12Bracket segments reported
0Individual identification
0Offshore data

The network could sell a digital screen with a defensible audience number: 18–34, male, dwell 4.2s, 9am–5pm, aggregated. Advertisers could brief against it. The network did not have to defend a mobile-location methodology.

Canopy Console — Screen audience 1 screen · 12 segments aggregated · no individual identification Audience index by hour 0 50 100 6:00 10:00 14:00 18:00 22:00 9am–5pm sell window Reported segments (12) 18–24 F M 25–34 F M 35–44 F M 45–54 F M 55–64 F M 65+ F M 6 age brackets × 2 · dwell ≥ 2s · median 4.2s
Audience console: hourly index plus the 12 aggregated bracket segments.

Implementation details

Each digital screen ran a small camera mounted above the display, feeding a Canopy CU-1 in the equipment cabinet. The audience model counted people who looked at the screen for more than two seconds, estimated a broad age/gender bracket, and discarded the frame. The event was aggregated: 18–34 male, 4.2 seconds dwell, 10am–11am, screen ID. No individual was identified. No data left Australia.

The network's sales team used the aggregated numbers to produce a defensible audience report for each screen. Advertisers could brief against a real number: how many looked, for how long, in which bracket. The report replaced the modelled traffic counts that had been the industry standard for a decade.

Screen portfolio

The operations team used the per-screen numbers to identify screens with low dwell and low audience, and renegotiated rent with landlords on that basis.

Ongoing support

The deployment now covers over 200 screens across the east coast. We add new screens to the network on a rolling basis. The audience report is produced automatically every week. Advertisers get the weekly audience number; the network gets a measured basis for selling screens.