Footfall Analytics: Visitor Counts, Dwell Time and What They Cannot Tell You

How footfall is counted

Every footfall number is only as good as the device at the door. The accuracy questions to ask before trusting any ratio: are staff excluded, are groups counted as people or as one, are children counted, and does the counter see both doors.

• Break-beam sensors: an infrared beam across the door. Cheap, and they count a group as one and cannot tell in from out.
• Thermal and 3D stereo counters: overhead sensors that count people and direction; the common choice for accuracy at a single entrance.
• Wi-Fi and Bluetooth sensing: counts phones probing for networks. Widely sold, and phone makers' address randomisation has made it less reliable than vendors' earlier claims, so treat it as a trend rather than a count.
• CCTV with computer vision: uses existing cameras, counts and tracks people across zones, and can exclude staff by zone or uniform. See the video analytics guide for the detail.

Visitor analytics: from a count to a pattern

A count becomes useful when it is read against time and against other counts. Visitors by hour and day show when the floor is busiest and whether the roster matches. Visitors by week against last year show whether the catchment is growing or the mall is fading. Where a camera or a loyalty scan can distinguish returning visitors, in aggregate and without identifying anyone, visitor analytics adds new against returning, which says whether marketing is bringing people in or the store is bringing them back.

Read visitors against staffing first. The most common finding in an Indian store network is a peak on Saturday evening with the same roster as Tuesday morning, which produces a queue, an unattended visit and a lost sale that no report shows.

Dwell time analytics

Dwell time is how long a visitor stayed, in the store or in a zone. Store dwell separates the bounce, in and out in under a minute, from the visit. Zone dwell, from cameras or sensors, shows where visitors lingered.

Dwell is ambiguous on its own. A long dwell at the counter is interest or waiting; a long dwell in a category is browsing or being unable to find help. Read it with the approach time and the bill, or with the conversation, before treating it as a good or bad number.

Footfall metrics: definitions, sources and owners

What footfall cannot tell you

A counter records the door. Everything a retailer wants to know about the visit happens after it. Three things in particular are invisible to footfall analytics and decide most of the revenue in an assisted store.

1. Unattended visits: the visitor who walked the floor, was never approached and left. The counter counted them in and out and reported a conversion loss with no cause.
2. Wait time: how long the visitor stood before an advisor reached them, and how many left during that wait.
3. The reason for leaving: what the attended visitor asked for, heard, and decided against.

The unattended visit, measured

The simplest measure of the unattended visit is the gap between footfall and conversations. Where advisors record consented sales conversations on their phone, coverage, conversations as a share of walk-ins, is reported by store and week, and the remainder is the share of visits that never became a conversation or were never recorded. Borentis reports coverage next to every score for that reason, and a store where footfall rises while coverage falls has a staffing or engagement problem that no counter would have named.

The fuller version joins the camera to the conversation: the approach event with its wait time, and the words that followed. ShopperDNA, which joins camera events to consented conversations, is on the roadmap. Until then, footfall read with coverage is the practical way to see the visits the counter counted and the floor missed.