Some numbers about a store are easy to trust and easy to act on. How many people walked in is one of them. How long they stayed, and where that time went, is harder to obtain and often more useful, because it is the closest thing a physical store has to a session on a website.
A shopper who enters, crosses to the back wall, spends four minutes comparing two products, and leaves without buying is a different problem from a shopper who enters and immediately walks out. A footfall counter records both as one visitor. Dwell time analytics is the layer that separates them — and, more importantly, tells you which part of the store produced those four minutes.
What dwell time actually measures
Dwell time is the length of time a person remains inside a defined area. Nothing about that definition is automatic, so three things have to be fixed before the number means anything:
- The area. A product display, an aisle, a fitting room, a checkout queue, and the whole sales floor are five different measurements with five different uses. Comparing dwell between two stores only works if zones are defined the same way in both.
- The start and the end. Most systems stamp the moment a tracked person enters a zone and the moment they leave it, then report the difference in seconds. People outside the zone are not counted, and anyone standing exactly on the boundary will flicker unless boundary handling is configured properly.
- The statistic. Average dwell is easy to read and easy to distort by a handful of very long stays. Median dwell describes the typical visit. Reporting both, plus the share of visits longer than a threshold such as two minutes, gives a far more honest picture than a single mean.
Store-level dwell — total time from entry to exit — and zone-level dwell are routinely confused. They answer different questions. Store dwell tells you how much time you had to sell. Zone dwell tells you where that time went.
Dwell time against the metrics around it
| Metric | Question it answers | Typical use |
|---|---|---|
| Footfall | How many entered? | Conversion denominator, staffing |
| Dwell time | How long did they stay, and where? | Display performance, engagement |
| Traffic flow | Where did they walk? | Layout, aisle design |
| Queue length and wait | How long did they wait? | Service level, checkout staffing |
| Conversion | How many bought? | The outcome everything feeds |
Dwell on its own is not a KPI. It becomes one when it is tied to an outcome. A display with above-average dwell and a rising attach rate is working. High dwell in a queue is a service failure. High dwell at a shelf with no sales usually means the product is out of stock, mispriced, or hard to find — not that shoppers love it.
How each sensor technology handles dwell
Dwell is a tracking problem, not only a counting problem. A device that counts people crossing a line may not be able to follow them afterwards.
- Infrared beam and threshold counters produce counts only. They cannot tell you how long anyone stayed, because they only see a break in a beam.
- 2D AI cameras count entries and, with zone polygons configured, can measure how long someone remains in an area. Performance depends heavily on camera angle and on how dense the store gets. Occlusion is the dominant failure mode: two people crossing closely can be merged into one track, which removes time from the count.
- 3D stereo (binocular) cameras add depth, which stabilises tracking where people overlap — the situation a busy aisle creates most of the day. Depth also makes the "did this person leave the zone, or were they just hidden behind someone else?" question much easier to answer.
- Ceiling-mounted panoramic cameras cover a large area from one position, which matters because dwell measurement needs continuous coverage. If a shopper leaves the frame and returns, most tracking software counts two visits unless the zones overlap cleanly.
- Thermal sensing reads body heat rather than visible light and keeps working in dark aisles and dim corners, though it produces less detail about what a person is doing.
Camera choice usually follows the ceiling. A three-to-four-metre mounting height over an aisle, a clear downward view, and shelving that does not block lines of sight will do more for dwell accuracy than any processing upgrade.
Five decisions dwell data changes
1. Display and planogram evaluation. Put a defined zone around a display and track median dwell alongside the conversion of the products on it. If people stop and do not buy, the display attracts but does not persuade.
2. Fitting rooms and high-consideration areas. Long dwell in a fitting area is expected. Dwell that collapses after a layout change is a warning that shoppers are no longer finding the items they came to try.
3. Floor staffing. If dwell spikes in an assistance zone without a matching increase in transactions, shoppers are probably looking for help. That is a scheduling signal, not a merchandising one.
4. Friction detection. Dwell rising near the front of the store while footfall stays flat usually means the checkout area is backing up. Pair it with queue measurement to confirm.
5. Campaign measurement. A promotion runs for two weeks. Compare zone dwell and conversion in those two weeks against the two weeks before — not against one quiet Tuesday.
The traps that make dwell data lie
Staff and contractors. Employees who stand in a zone for an hour will dominate a median. Either exclude their tracks in the analytics configuration, or use a device that deliberately does not maintain long-term identity so their influence stays bounded.
Long dwell is not automatically good. Waiting for a companion, hunting for a price tag, and standing in a queue all look like engagement, and none of it sells anything. Dwell only becomes insight once you know which of these is happening.
Small samples. A zone that receives fifteen people a day produces a median that swings wildly week to week. Aggregate over a longer window, or compare only zones with similar traffic.
Boundary flicker. Badly drawn or overlapping zones make one person generate several short visits, dragging the average down and inflating visit counts. Review the drawing with real traffic on the floor, not on an empty morning.
Comparing unlike things. Dwell in a supermarket aisle and dwell in a boutique fitting area differ for structural reasons. Benchmarks are most useful within your own chain, across stores with similar layouts.
A practical first deployment
- Pick five zones that correspond to decisions you already make — not a detailed map of the entire store.
- Measure for two weeks without changing anything. You need a baseline before you need a plan.
- Tie one zone to one outcome: transactions for a promoted item, service calls, or conversion of the fitted area.
- Change one thing — move the display, adjust a shift, rework shelving.
- Compare against the same window in the previous cycle. If the change does not show up in the numbers, it was not a change.
Dwell data is most valuable as a routine, roughly monthly review of a handful of zones. A dashboard nobody opens is worse than no dashboard at all.
Dwell time by store format
The same metric behaves differently across retail formats, which is why cross-format comparison misleads.
Grocery and convenience. Visits are short and purposeful, and dwell concentrates around fresh produce, promos, and the checkout run. Useful zones are few and change with the weekly promotion, so zone definitions need maintaining rather than setting once.
Fashion and footwear. Dwell is long by nature, especially around fitting rooms and mirrored areas. The interesting number is not total time but the ratio of display dwell to fitting-room dwell — shoppers who browse for ten minutes and never try anything are a sizing, signage, or price problem.
Electronics and home appliances. High consideration, high assistance need. Dwell at a demo counter with no sales assistant in range is one of the clearest staffing signals a store can produce, and it is invisible in a footfall report.
Pharmacy and small-format specialty. Narrow aisles make occlusion worse and zones smaller. Plan for a steeper overhead view and expect lower confidence in any single zone; aggregate across the week rather than reading a single day.
Banks, clinics, and service counters. Dwell is dominated by queueing, which is why these venues need queue analytics as a separate layer. Long dwell there is a service failure, not engagement.
The practical rule is to compare a store against itself over time, or against stores of the same format, and to treat any figure labelled as an industry benchmark with the scepticism it deserves.
FAQ
What is a good dwell time for a retail store? There is no universal benchmark, and anyone quoting one without knowing your format is guessing. Grocery, fashion, electronics, and pharmacy stores have structurally different dwell patterns driven by basket size and store size. Use your own baseline and your own other stores as the comparison.
Is dwell time the same as time spent in the store? No. Time spent in the store is a single number per visit. Zone dwell breaks that number into parts, which is what makes it actionable: it tells you where attention was spent, not only how much of it there was.
Can dwell time be measured without cameras? Only roughly. WiFi and beacon systems can estimate how long a device stayed in an area, but they rest on assumptions about who carries a phone and how many devices one person has. That is reasonable for trends and weak for absolute figures.
Does dwell tracking identify individual shoppers? A properly configured retail system stores anonymous tracks — positions and timestamps, not identities. The limits are practical as well as legal, and how long data is retained should be a written decision rather than a software default.
---
Footfall tells you whether people came; dwell tells you whether the visit was used well. The solutions overview describes how zone and dwell analytics fit retail and public-space deployments, and the product center lists the camera families built for overhead and zone coverage. For how dwell connects to the metrics around it, see the blog. If you have a floor plan to work from, contact our team.