How to Audit a Footfall Claim Before You Trust It

Knowing how to audit a footfall claim is a survival skill for anyone running a street-front business. Landlords quote footfall to justify rent, data vendors quote it to sell subscriptions, and marketing decks quote it to win your budget — and most of the time nobody hands you the working behind the number. This guide gives you a short, non-technical audit you can run on any footfall claim in a couple of minutes, plus the one blind spot that trips up even expensive data: what happens at the level of a single window.

It builds on why you should never trust an unverified number. That post makes the case; this one is the method.

How to audit a footfall claim in five questions

Run any footfall claim through these five questions. You are not looking for perfection — you are looking for whether the number can defend itself.

1. What was counted, and where? People or devices? On your pavement, or across a whole area averaged out? A claim that cannot point to the exact line it counted is describing a neighbourhood, not your door.
2. How big was the sample, and over what period? One hour, one day, a week? Weekday or weekend? Sunny or wet? The narrower the sample, the less it travels to other days.
3. What is the margin of error? A single number with no range is hiding its uncertainty. A trustworthy claim states one; here is how to read those ranges.
4. Can you see the evidence? Raw hourly counts, an annotated clip, a method note — something you can actually inspect. "Our model says so" is not evidence you can check.
5. What does it leave out? Dark hours, blocked views, dense crowds, one-sided counting. An honest claim names its own limits; a sales figure hides them.

If a claim answers all five, it has earned real weight. If it dodges two or more, downgrade it to "interesting, unproven" and do not build a budget on it.

The blind spot: panel data at a single window

The most respectable-looking footfall claims often come from mobile location panels — services that estimate an area's footfall from a sample of smartphone-app users, then scale that sample up to a total. This is genuinely useful technology for the job it was built for: comparing districts, spotting seasonal trends, ranking retail zones across a city. If you want to know how Baixa compares to Cedofeita, a panel is a reasonable tool.

But panels have a structural blind spot at the scale you care about — your specific window, on a specific day. Here is why:

• They are modeled, not counted. A panel sees a fraction of phones and extrapolates. The scaling assumptions are fine over a big area and shaky over a few metres of pavement.
• They are zone-level. The unit is a place or block, not the six metres of frontage you pay rent on. Two shops on the same block can have very different pavements, and a panel cannot tell them apart.
• They cannot see capture. A panel might estimate how many phones were near your street. It has no idea how many people stopped at your window or came through your door — so it can never give you a capture rate.

This is the core of panel versus ground truth. Ground-truth counting watches video of your exact frontage and counts each person who crosses your line. It is narrow — one spot, the period you filmed — but for that spot it is the real thing, not a model. For a single shopfront decision, narrow-and-real beats wide-and-modeled every time.

Key takeaways

• Audit any footfall claim with five questions: what/where, sample/period, margin of error, evidence, and limits.
• Panel data is modeled and zone-level — strong for comparing districts, blind at a single window.
• Only counted footage of your exact frontage can give you a capture rate; panels cannot see your door.
• A claim that dodges two or more of the five questions is interesting, not proven.

Auditing your own numbers, not just other people's

The same five questions apply to the report in your own hands — and a good audit is built to pass them. A Capture Rate Audit answers "what was counted, where" with your line placement, "how big a sample" with the observed period, "margin of error" with a confidence interval, "evidence" with a 60-second annotated clip, and "what it leaves out" with an honest limits note when the light or the crowd made a stretch of footage hard. Reading the audit walks through each of those in turn.

That symmetry is the point: we hold our own numbers to the standard we are asking you to hold everyone else's. We count silhouettes, not people — no faces, no identities — and delete source video after processing.

Put a claim to the test

If someone has handed you a footfall figure, the cheapest way to audit it is to measure the same spot yourself and compare. Get a $99 Capture Rate Audit for your frontage and hold it up against the claim, or see pricing for a seven-day audit that settles the question. And if you are opening a second site rather than auditing your current one, that is a pre-lease decision — our sister brand handles it at streetproof.net.

Frequently asked questions

How do I audit a footfall claim? Ask five questions: what was counted, how and where; how big the sample and period were; what the margin of error is; whether you can see checkable evidence; and what the number leaves out. If the claim cannot answer them, treat it as indicative, not fact.

What is the difference between panel data and ground-truth counting? Panel data models an area's footfall from a sample of mobile-app users, then scales it up. Ground-truth counting counts each person who actually crosses a line in video of the exact spot. Panels are strong for wide-area trends and blind at a single window; counting sees the specific pavement.

Is mobile location footfall data wrong? Not wrong — just built for a different job. It estimates area-level patterns well, but it is modeled and extrapolated, so it cannot tell you how many people passed your specific window this Tuesday. For a single shopfront, counted footage is the ground truth.