Measurement
Part of Out of home advertising measurement works when the method matches the decision
Why out of home advertising attribution methods need a decision first
A comparison of out of home advertising attribution methods, matching exposure modelling, control groups and logged responses to the decision each can support.
What to take away
- Start from the decision, not the metric. Route 60 of the UK's major roads carried 75.3 billion vehicle miles in the year ending June 2025, according to Department for Transport statistics published in 2025. Exposure is easy to model and hard to tie to a sale.
- Exposure-based methods such as footfall and traffic counts are cheap and replicable, but they measure opportunity to see rather than response.
- Control or holdout designs give the closest thing to a causal read, but they need scale, a clean comparison group and time.
- Logged responses such as codes, tracked URLS and QR scans are precise for the people who act, and silent about everyone else.
- Agree the reporting line before buying, using the structure set out in out of home advertising measurement and reporting.
Which attribution method answers which question?
No single method covers every question, so buyers run two or three together. Each produces different evidence, and each suits a different decision.
| Method | Evidence it produces | Best for | Main weakness |
|---|---|---|---|
| Exposure modelling | Estimated impressions and reach | Comparing sites before buying | Says nothing about behaviour |
| Control or holdout test | Difference in outcomes between exposed and unexposed areas | Causal claims for a campaign | Needs scale and a clean control |
| Logged response | Codes, tracked URLS and scans traced to a source | Direct response and promo activity | Captures only those who act |
| Geo lift or sales modelling | Sales or footfall change in treated catchments | Retail and local reviews | Model assumptions and data quality |
Most buys need one planning method and one proof method running side by side, and the plan should price them separately.
How should you choose between exposure and response measures?
Exposure measures suit planning decisions. Comparing two roadside formats means weighing modelled reach and cost per thousand impressions like for like. Exposure figures are cheaper to produce and easier to compare across sites.
Response measures suit trading decisions. If the question is whether a promotion worked, a tracked URL or code answers it for the people who used it.
Coverage is the limit. Most passers-by never interact, so response counts understate the total audience and flatter the channel.
Pair one exposure measure with one response measure, then state in the plan which decision each informs. Label both in every report so nobody reads a modelled figure as a measured one.
What does a control or holdout design add?
A holdout removes advertising from matched areas or times and compares outcomes with treated ones. Done well, it is the closest thing to a causal read most campaigns can get.
It is also demanding. You need enough sites, a comparable control group and time for the effect to show. A ten-site test with two control sites will not settle a national question.
Where a holdout is impossible, a stepped regional rollout can approximate it. Record the design in advance, including the primary outcome and the analysis window.
Do digital screens change the attribution question?
Digital out of home adds scheduling flexibility and, in some cases, online delivery alongside physical screens. Google's explanation of the Display Network describes how ads are matched to sites and apps, useful context when a screen's inventory also appears in digital buying.
Even so, a digital screen is still seen in a place, at a time, by people who may never click. Treat online delivery data as one input, not proof that the physical placement worked.
Before any campaign that uses location or device data, check the supplier's record on data protection. The ICO's published decision notices show how breaches of data rules are handled in practice, useful due diligence material for a data-driven attribution supplier.
What should the reporting show?
A useful report separates what was bought, what was observed, and what is inferred. Buyers who blur those three lines cannot defend the number when finance asks. Name the method behind each figure, not just the result.
Set the reporting cadence to the decision cycle. Weekly trading reviews need fast response counts, while quarterly reviews can carry modelled reach and test results.
A standing out of home advertising reporting dashboard keeps those layers visible and stops the evidence being rebuilt for every meeting.
Claims made in out of home creative fall under the same advertising rules as other media. Check the ASA's news and updates for rulings touching out of home.
Common questions
Can one method cover both planning and proof?
No. Exposure modelling answers planning questions, and control tests answer causal ones. Run both, and label which is which.
How long should a holdout run?
Long enough for the outcome to move measurably, which depends on the category and the number of sites. Agree the window before the campaign starts.
Are modelled impressions acceptable evidence?
They are acceptable for comparing and pricing sites, but weak as evidence of behaviour change.
What should a supplier provide for due diligence?
Method notes, data sources and dates, plus compliance evidence where personal data is used.
