How do location intelligence platforms measure footfall from a pDOOH campaign? They use location data from a panel of mobile devices to find devices that were near a screen while your ad played, then compare how many of those devices later visited your store with a matched group of devices that were not exposed. The difference is the footfall lift. Done well, it is a useful estimate of what the campaign added. Done badly, it can credit the screens with visits that would have happened anyway.
This guide explains the method step by step, where it can go wrong, how to read a lift figure with a worked example, and what to ask a provider before you rely on the result.
How a footfall study works
- Define exposure. The platform takes the play logs for your campaign, showing which screen played your creative and when, and draws an exposure zone around each screen. The OAAA's attribution best practices say play logs are required for digital OOH so exposure is tied to the creative actually shown.
- Find exposed devices. Devices in the location panel that were inside an exposure zone while your ad played form the exposed group. The panel is made up of phones whose apps share location data with consent.
- Define the visit. A boundary is drawn around each store. The MRC, IAB and MMA location-based measurement guidelines say visits should be validated with rules for speed and dwell time to separate real visits from drive-bys, and that multi-storey buildings need altitude data.
- Build a control group. Unexposed devices are matched to the exposed group so that, as far as possible, exposure is the only difference. The OAAA guidance says the control must never simply be every unexposed device, and should match on likelihood to visit.
- Compare visit rates. Over a set window after exposure, the platform compares the share of each group that visited. The excess in the exposed group is the lift.
For how footfall fits alongside other outcome measures, see our guide to measuring pDOOH success.
Where a footfall study can go wrong
| Weak point | Why it matters | What to check |
|---|---|---|
| Location accuracy | The MRC guidelines put GPS at roughly 10 to 100 metres and note it degrades indoors; Wi-Fi at 30 to 100 metres; cell towers are coarser | Which signals are used, and the accuracy and precision the provider discloses |
| Shrinking panels | Fewer devices share location data, so panels get thinner | Panel size in your market and how many exposed devices the study found |
| Presence is not attention | A device in the zone does not prove the person saw the screen | How the exposure zone is defined for each screen |
| Weak control group | A poorly matched control inflates or hides lift | How the control was matched, and whether pre-campaign visit rates were similar |
| Attribution window | Too short misses effects; any window can pick up other activity | Window length; the OAAA guidance suggests at least 30 days |
| Other media | A single-channel study can credit OOH with the effect of TV, search or promotions | What else ran in the period, and how the report handles it |
Perion, which sells DOOH through Hivestack, makes the same point about thin data in its guide to DOOH footfall measurement: as opt-in to mobile advertising IDs declines, panels shrink, and results based on thin data are unreliable however they are presented. On iPhones, Apple's App Tracking Transparency rules require apps to ask permission before tracking users across other companies' apps and sites, which is one reason panels have shrunk.
Illustrative example: reading a lift figure
Illustrative example. All figures are hypothetical, and the scaling step relies on an assumption that a real study would need to justify.
- The study finds 200,000 exposed devices, of which 1.8% visited a store within 30 days. A matched control of 200,000 devices visited at 1.5%.
- Absolute lift: 1.8% − 1.5% = 0.3 percentage points. Relative lift: 0.3 ÷ 1.5 = 20%.
- Incremental visitors in the panel: 200,000 × 0.003 = 600.
- Assume the panel represents 5% of devices in the area. Scaled up: 600 ÷ 0.05 = 12,000 incremental visitors.
- With $30,000 of media, that is $30,000 ÷ 12,000 = $2.50 per incremental visitor.
Now check whether the difference is real rather than noise. The standard error of the difference between the two rates is about 0.04 percentage points with 200,000 devices in each group, so a 0.3-point gap is far outside chance. With 10,000 devices in each group, the same rates give a standard error of about 0.18 points, and the gap is no longer significant at the usual 95% level. The same headline lift of 20% can be solid or meaningless depending on sample size, which is why the OAAA guidance asks whether matching leaves enough data for a statistically significant result.
Two more cautions: the scaling assumption drives the cost per visitor, so ask how it was derived, and "relative lift" sounds larger than "absolute lift" for the same result, so ask for both.
Privacy and the rules on location data
Footfall studies depend on precise location data, which regulators treat as sensitive. In the US, the Federal Trade Commission finalised an order against InMarket in May 2024 that bars the company from selling or sharing precise location data, after alleging it had not ensured consumers gave informed consent. Under GDPR in Europe and similar laws elsewhere, the same questions apply: how consent was collected, what is retained, and whether sensitive places are excluded.
For buyers, the practical step is to ask the provider where its panel data comes from, how consent is obtained, whether sensitive locations are filtered out, and whether results are only reported in aggregate. A provider that cannot answer clearly is a risk to your brand as well as to your results.
Questions to ask a footfall provider
- Which location signals do you use, and what accuracy and precision do you disclose?
- How large is your panel in our market, and how many exposed devices did this study find?
- Do you use our play logs to define exposure, and how is each screen's exposure zone drawn?
- How was the control group matched, and were pre-campaign visit rates similar?
- What attribution window did you use, and why?
- Can you report absolute lift, relative lift and confidence intervals?
- How do you scale panel results to the population?
- What other media and promotions ran, and how does the report account for them?
- How is consent collected, and are sensitive locations excluded?
Using footfall data well
Footfall results are most useful for comparing options within your own campaigns, such as which screens, venues or dayparts produce more lift, rather than as an absolute verdict on the medium. Some DSPs can use these results to shift budget during a campaign; it is worth asking about when you are choosing a DSP, and it is the kind of automated optimisation discussed in our guide to AI-powered pDOOH. In-store screens add another option, measurement against the retailer's own sales data, covered in retail media networks and programmatic DOOH.
Footfall data is only as good as the delivery behind it, so pair it with the checks in our guide to ad verification in programmatic DOOH. Buying reach in the wrong places is one of the common programmatic DOOH campaign mistakes that a good footfall study can expose. For context on how footfall fits into return on investment, see our pDOOH versus traditional billboards ROI comparison, the latest programmatic DOOH statistics, and real programmatic DOOH campaigns worth studying.