Measurement
Part of Out of home advertising measurement works when the method matches the decision
Planning an out of home advertising reporting dashboard
An out of home advertising reporting dashboard framework for campaign decisions, delivery and reach metrics, data sources, scoring and audit trails.
What to take away
- Start from the decision the buyer will make at the next review. A dashboard for a Manchester bus shelter campaign should answer one question: renew, shift budget, or stop.
- Report counts and reach together. Poster sites deliver impressions, but a retail client cares about footfall near the store, so pair OTS figures with a location measure.
- Keep every number on the out of home advertising dashboard traceable to a source, a date and an owner. Untraceable figures do not belong on the dashboard.
- Score each metric against a written rubric before launch, so nobody argues about weighting after results land.
Decide the decision before the chart
A dashboard fails when it is built backwards from available data. The useful order is decision, metric, source, then visual.
Decision-first build order
- Decision: what action follows
- Metric: number for that action
- Source: where data comes from
- Visual: chart built last
Ask the buyer what action follows the report. For a six-sheet England roadside campaign, the likely actions are extend, reweight sites, or reallocate to another format. Each action needs a different number.
Write the decision at the top of the dashboard file. It keeps the build honest when someone asks for one more chart. Keep the link between evidence and the buyer’s choice in view, rather than presenting figures without a decision.
Choose metrics that survive scrutiny
Counts are the base layer: plays, impressions, and site days delivered against site days booked. Reach and frequency sit above them, usually modelled rather than observed.
Metric layers in OOH
Counts
- What it is
- Base layer
- Examples
- Plays, impressions, site days
- Best for
- Delivery checks
Reach and frequency
- What it is
- Modelled layer
- Examples
- Reach, frequency
- Best for
- Audience scale
Location measures
- What it is
- Context layer
- Examples
- Drive-time, footfall bands
- Best for
- Grocery catchment fit
Location measures sit alongside. For a grocery client, a drive-time or footfall band around each site often matters more than gross impressions.
If the campaign uses dayparting, mirror the planning schedule logic: the dayparts agreed in the plan drive the time bands on the dashboard, so commuter peaks and evening retail windows are reported as they were bought.
Build the dashboard in four steps
- List the decisions from the kick-off meeting and CAP the list at five.
- Map each decision to one primary metric and no more than two supporting metrics.
- Record the source, refresh date and owner for every field in a data dictionary.
- Draft the layout on paper, then build only what the paper version needs.
The data dictionary is the part teams skip. It is also the part that saves a review meeting when a number is challenged. A first draft usually lists campaign, site ID, format, week, plays booked, plays delivered, impressions, modelled reach, modelled frequency, footfall band and cost per thousand.
Decide the build platform before the layout. Looker Studio, Power BI and Tableau can all carry the four-step structure above, and the choice usually comes down to who owns the licence, who can refresh the data and whether the client already works in one of them.
Where the data comes from
Route counts, play logs and site schedules come from the media owner: for roadside and transport formats in Great Britain that usually means JCDecaux, Clear Channel UK, Global or Ocean Outdoor, and Outsmart is the trade body those owners work through. Ipsos conducts Route, the audience measurement system for the out-of-home advertising industry in Great Britain, for the industry body with the same name, which is where the modelled audience figures behind reach and frequency come from. Post-campaign research comes from a fieldwork agency briefed on the same site list.
Label modelled figures as modelled. A modelled reach figure is an estimate, not a census, and the dashboard should say so next to the number.
How to keep it auditable
Version every refresh. Note who changed what and when. If a client asks why reach moved between weeks, the answer should be in the log, not in someone's memory.
Score results against a rubric
A rubric turns a pile of metrics into a defensible verdict. Agree it before launch and keep it short.
| Criterion | Weight | What good looks like |
|---|---|---|
| Delivery against booked site days | 25% | Within 5% of plan |
| Reach against target audience | 25% | Meets or beats the agreed threshold |
| Location fit | 20% | Sites sit inside the agreed catchment bands |
| Cost per thousand delivered | 15% | At or below the planned rate |
| Data completeness | 15% | No unexplained gaps in the refresh log |
Weights are illustrative. A brand campaign may shift weight towards reach; a retail campaign may shift it towards location fit. Agree thresholds before launch and document the rationale, so the scoring is not changed after results land.
Score each criterion out of five, multiply by the weight, and total. A score below the threshold triggers a review, not automatic cancellation.
Handle privacy and complaints properly
Any dashboard that ingests movement or location data needs a lawful basis and a data protection impact assessment under the UK GDPR and the Data Protection Act 2018. The ICO's guidance on designing products that protect privacy sets out the privacy by design principles that apply to digital OOH products.
Complaints about a creative or a site sometimes end in an ASA ruling. If a party disputes the outcome, the independent review process for ASA decisions explains how that challenge works.
Keep complaint records and ruling references in the dashboard archive. They are evidence for the next buy.
Common questions
How many metrics should a dashboard show?
Aim for five to seven headline metrics tied to decisions. More than that and the review becomes a reading exercise.
Should modelled reach sit next to observed counts?
Yes, but label them differently. Observed counts and modelled estimates carry different confidence, so unlabelled mixing misleads the reader.
How often should the dashboard refresh?
Match the refresh to the buying cycle. Weekly suits short flights; monthly suits long-format tenancies. A refresh nobody reads is wasted effort.
What belongs in the archive?
Source files, the data dictionary, the rubric and any complaint or ruling correspondence. An archive that cannot be reconstructed six months later is not an archive.
