Video and vision
How to Measure Customer Movement Using CCTV: Footfall, Dwell, Paths and Zones
To measure customer movement using CCTV you draw lines and zones on each camera's view and count what tracked people do against them: crossing the entrance line is footfall, time inside a zone is dwell, the sequence of zones a track visits is a path, and buyers divided by visitors in a zone is zone conversion. Each is a real measure with a known error. The purpose is not the numbers themselves but the decisions they support: staffing by hour, display placement, and the moment a shopper went unattended.
The four movement measures
Footfall counts people crossing a threshold, usually the entrance, with direction. Dwell times how long a track stays inside a zone. Paths record the order of zones a track visits. Zone conversion divides transactions attributed to a zone by the tracks that dwelt there. The first two run on ordinary security cameras. Paths need overlapping views. Zone conversion needs a POS join and a rule for which zone a sale belongs to, which is honest only in stores where product categories have distinct floor areas.
Footfall analytics at the door
- One camera inside the store facing the doorway, high enough that people do not overlap as they enter.
- A line across the threshold with an in and out direction; a second line a metre inside catches shoppers who step in and step out.
- Staff, delivery and security excluded by a staff zone or a schedule rule.
- Groups: tracks that cross within a couple of seconds and stay together are counted as one group and n people, both kept.
- Validation: a hand count for two hours on a weekday and a weekend, repeated after any camera change.
Dwell time analytics by zone
- A zone is a polygon on the image, not on the floor plan; each camera has its own.
- Dwell starts when a track's foot point enters the polygon and ends when it leaves; a minimum threshold, often 10 to 20 seconds, separates a dwell from a pass-through.
- Report median dwell and the share of visitors who dwelt, not the mean, which one long-standing shopper distorts.
- Crowding shortens measured dwell because tracks break and restart; read dwell with the footfall of the same hour beside it.
How each measure is computed and where it fails
| Measure | How it is computed | Main error source | How to check |
|---|---|---|---|
| Footfall | Directional line crossings by tracks | Re-entries, staff, door groups | Hand count, two hours, two days |
| Dwell time | Time a track's foot point spends in a zone | Track breaks under crowding, threshold choice | Follow ten shoppers by eye per zone |
| Heatmap | Accumulated track positions per cell | Perspective bias from wall mounts | Compare to a top-down camera once |
| Path | Ordered zone visits per track across cameras | No handover between views; ID switches | Only trust within one camera's view |
| Zone conversion | Attributed sales divided by zone visitors | Attribution rule, shared counters | Reconcile to POS weekly |
| Queue time | Duration of tracks in the counter zone | Browsers near the counter | Time five queues by phone |
| Time to first contact | Gap between entry and first staff proximity | Staff exclusion failures | Sample twenty visits by eye |
Turning movement into decisions
- Staffing: plot unattended visits and queue time by hour; put the extra advisor on the hours, not the day.
- Layout: a zone with high dwell and low zone conversion is a display that attracts and does not sell; a zone with low dwell and high conversion is under-promoted.
- Entrance: a high step-in-step-out count is a window or an entrance display that promises something the first metres do not deliver.
- Standards: set time to first contact as a store standard and report the hours it slips.
Movement is the map, not the reason
Movement measures describe where shoppers were and for how long. They cannot say why the shopper who dwelt four minutes at the premium range walked out, or what the advisor who reached her after ninety seconds actually said. The camera maps the space around the sale; the consented conversation records the sale. Read the two by the same hour and zone and the map gets its reasons. There is no need to know which shopper was which, and no lawful way to find out.
Borentis captures consented in-store conversations on the advisor's phone today and scores them on the playbook. ShopperDNA, on the roadmap, would add footfall, dwell, wait and walk-out events from a store's existing CCTV and join them to those conversations by time and zone.
Frequently asked questions
How accurate is CCTV footfall counting?
Within a few per cent on a well-placed entrance camera after a week of validation. Wall-mounted cameras that see the door at a slant do worse, and doors with groups arriving together need a group rule.
What is a good dwell time in a store?
There is no universal number. Compare a zone to itself week over week and to other zones in the same store. A dwell that rises while conversion does not is the useful signal.
Can CCTV track a customer's whole path through the store?
Only with overlapping, near top-down cameras. On typical security mounts, paths are reliable within one camera's view and break between views.
Do I need a people counter if I have CCTV analytics?
Not usually. A dedicated overhead counter is more accurate at busy doors; an entrance camera with a validated line is good enough for most showrooms and EBOs.
Related reading
- How to use CCTV for retail analytics
- How AI detects customer behaviour in stores
- Conversation intelligence vs CCTV and footfall analytics
- Why CCTV alone cannot explain retail conversion
Where Borentis applies this
- Walk-in Recovery: The customer who left is still yours.
- Execution Scorecards: See the floor before the P&L does.
Borentis is the Agentic Operating System for Customer Interactions, built for Indian retail floors: consented one-tap capture on the advisor's phone, every conversation scored against your playbook with the evidence behind every number, leads created when a number is heard, and coaching from your own best conversations.