Proof of play: how to measure mobile DOOH campaigns
An agency guide to verifying mobile DOOH: play logs, GPS and timestamps, impression modelling, data feeds and what to ask a vendor before you buy.
Mobile digital out-of-home comes with an obvious question. If the screen is moving, how do you know your ad ran where you paid for it to run?
The answer is proof of play. For agencies, it decides whether mobile DOOH is a format they can report on with confidence or one they have to take on trust. This guide covers what a mobile proof of play record should contain, how impressions are estimated, how to get the data into your own systems, and what to ask a vendor before you commit budget.
What proof of play means
In digital out-of-home, proof of play (often shortened to PoP) is a log generated by the screen or its player each time a piece of creative is shown. A fixed screen never changes location, so a log of what played and when is often enough.
Mobile screens add location to that record. The same screen can be in Soho at noon and Islington an hour later, so a useful mobile PoP record has to answer four questions for every single play: what ran (the creative and campaign identifiers), when it ran (a precise start and end timestamp), where the screen was (a GPS position at the time of play) and which screen showed it (a unique screen or device ID).
If a vendor cannot give you all four at the level of the individual play, you are relying on summaries you cannot check.
Anatomy of a play record
A well-formed mobile PoP record usually contains something like the following fields.
| Field | What it tells you |
|---|---|
| Play ID | A unique reference for this play |
| Screen ID | Which physical screen showed the ad |
| Campaign and creative IDs | Which ad and which version |
| Start and end timestamp | When it played, and for how long |
| Latitude and longitude | Where the screen was during the play |
| Zone | Which geo-fence the play counted towards, if any |
| Screen status | Whether the display was on and healthy |
| Impression estimate | Modelled number of people likely to have seen it |
With this much detail an agency can rebuild the campaign from raw data, check delivery against the plan and spot anything odd.
Where the data comes from
Telemetry
Connected screens report their status continuously, including their location, whether the display is powered and playing, and whether the content is up to date. This telemetry is what makes play-level proof possible. It also lets the platform stop serving zone-specific creative the moment a screen leaves a zone.
GPS
Satellite positioning gives each play a location. GPS can drift in dense city streets, especially near tall buildings. Good systems smooth the track and flag low-confidence positions instead of reporting them silently, so ask how a vendor handles this.
Player logs
The media player on the screen records exactly what it played and for how long. Matching the player log to the telemetry and the GPS track produces the play record.
From plays to impressions
A play is something that happened and can be logged. An impression is an estimate, and it is worth making that difference clear to clients.
Out-of-home impressions have always been modelled, because nobody counts every pair of eyes that passes a billboard. The UK out-of-home industry has its own audience measurement currency, Route. Mobile formats that go down smaller streets also need a model built around how they move.
At Curb the modelling works from where and when each play ran. The bags have no cameras and no facial recognition, so nobody on the street is filmed to produce the numbers.
How mobile impression modelling works
An impression estimate for a mobile play typically starts with where the screen was, down to street level, and when. Footfall on a Soho street at 1pm is very different from 4am. The model then brings in footfall and population data, which estimates how many people are on foot or nearby in that area at that time. It also accounts for visibility: how long the screen was in a position to be seen, and how fast it was moving.
The output is an estimated number of people who had the opportunity to see that play. Summed across a campaign, this gives total impressions, which can then be broken down by zone, daypart and creative.
Questions to ask about any impression model
- What data sources feed it, and how often are they refreshed?
- Is the estimate calculated per play or averaged across the campaign?
- How does it treat time of day and day of week?
- Is it an opportunity-to-see estimate, and does the vendor say so clearly?
- Has the methodology been documented so your team can review it?
A vendor who explains their method plainly, limits included, is more use to you than one quoting large numbers without context.
Getting the data into your stack
A dashboard is fine for a quick check, but agencies usually need more than that.
Live dashboard
A live view of the campaign shows plays, zones, impressions and screen activity as they happen. Planners can see delivery early and adjust if a zone is under-delivering.
Data feed or API
For proper reporting, agencies should be able to pull play-level data into their own systems. That might be a scheduled export or an API that returns plays, impressions and metadata for a campaign. With the raw data in hand you can join it to your other channels, apply your own attribution models and keep a record that does not depend on the vendor.
Curb provides a live advertiser dashboard, which can be filtered by zone, and an agency data feed, so play-level data can sit alongside the rest of a client’s reporting.
Attribution for mobile DOOH
Proof of play tells you the ad ran. Attribution asks what happened next. For out-of-home it is always indirect, but there are sensible ways to approach it.
Exposed versus control areas
Because every play is geo-located, you can compare zones where the campaign ran with similar zones where it did not. Look for differences in store footfall, app installs, branded search or sales over the same period.
Time-based lift
For dayparted campaigns, compare the hours the creative ran with the hours it did not. This works well for food and drink, where the gap between seeing an ad and ordering can be short.
Offer codes and landing pages
Area-specific codes or URLs give a direct signal, though they tend to undercount, because many people who respond will not use them.
Brand studies
Surveys in exposed and unexposed areas can measure shifts in awareness and consideration. They take longer to set up, but they answer the question most brands care about.
For more on planning the zones and dayparts that make exposed-versus-control testing possible, see our geo-fencing guide for London.
A checklist for agencies
Before booking any mobile DOOH campaign, ask for:
- A sample play-level log with timestamp, GPS and screen ID.
- A written description of the impression methodology.
- Access to a live dashboard during the campaign.
- A data feed or API for play-level data.
- An explanation of how GPS errors and screen faults are handled.
- Clarity on what happens if delivery falls short of the plan.
- Evidence of how the creative will look on the hardware. Display quality matters too, as we cover in LCD vs LED screens for mobile advertising.
Summary
Mobile DOOH can be as accountable as any other channel, provided the data is there. That means a timestamp, a location and a screen ID on every play, and impressions modelled per play with the method written down. Agencies should also be able to take the raw data away and use it in their own reporting.
If street-level media is new to you, What is street-level media? sets out how the format works. The advertisers page covers Curb’s measurement in more detail. Our media pack goes further on reporting and data access, and you can request it here.