People Counting And Store Analytics For Retail
Conversion rate needs a denominator. Your cameras already have it. ARSA turns existing store CCTV into footfall, queue, and dwell data without adding sensors at every door.
Counts people anonymously by default. Repeat visitors and staff recognition are opt-in, with consent.
Core Applications
Retail Footfall, Queue And Dwell Analytics On Existing CCTV
Most stores already film every entrance and till, and the footage is only watched after a theft or a complaint. Head office sees sales by store and has to guess the traffic behind them, so conversion, staffing and layout decisions rest on estimates. Door counters give a single total and say nothing about queues or which aisles people stop in. Video analytics reads the cameras you already have and turns them into counts, queue times and dwell by zone, for every store in the estate.
- Store camera
- AI Box in the store
- Counts on the dashboard
Footfall & Occupancy
Entry and exit counts per store and per entrance. Track real-time occupancy levels against safety or operational capacity limits using existing door and overhead cameras.
Queue & Dwell Analytics
Measure queue length and wait times at till points. Track dwell time by specific department or fixture zones, and generate heatmaps of aggregate floor movement.
Every Store In Your BI Stack
Send every store's counts, queues and conversion through the API and signed webhooks, or export them, into the BI reporting stack your merchandising and operations teams already use.
Challenges
What A Store Estate Rollout Has To Handle
Cost Per Store
A chain evaluates on the per-store number. One AI Box Mini per store costs $2,260 for the hardware plus $29 per camera per month for Retail Counter, so $87 a month for a three-camera store, billed annually. The arithmetic is different at 10 stores and at 400.
Existing Sensor Investment
Most chains already own door counters, and they can stay in place. ARSA adds the zone-level and queue data those sensors cannot produce, from cameras you already have.
Privacy Expectation
Retail is the sector where shopper identification is most likely to be assumed and most damaging to assume. State the position in the store and on the page: the system counts shoppers anonymously by default and keeps no face templates unless the store switches on repeat visitors or staff and VIP recognition.
Camera Placement For People Counting
A count is only as good as the camera’s view of the doorway or the till. Our camera standard, ARSA-CVS-001, sets the figures every store camera is checked against before a rollout:
- Cameras: 2 MP minimum, 4 MP recommended, at 15 fps or more.
- Entrances: an overhead or steeply angled camera above the door, 60 to 90 degrees below horizontal, giving at least 113 PPM (pixels per metre) at the counting line, with at least 50 lux of light.
- Queues at the till: an angled view of 30 to 50 degrees, 55 at most, giving at least 99 PPM along the queue, with at least 50 lux. With a standard 60 degree lens a 2 MP camera reaches 16.8 m and a 4 MP camera 22.4 m.
- Dwell and heatmaps: a wide, angled view that covers the whole department or fixture zone, checked camera by camera in the site survey.
- Camera settings: smart codecs off and a constant bit rate on the stream the analytics read.
Maximum distance with a standard 60 degree lens
An overhead camera sees the tops of heads, which suits a store that counts without identifying anyone. If one of your cameras falls short, the free AI assessment report says which one and how to move it; the full figures are on the system requirements page.
What It Costs Per Store, And What It Has To Earn
Take a store with three cameras, one at the entrance and two over the tills. Bought outright, an AI Box Mini costs $3,760 once, with perpetual counting and 12 months of updates, support and warranty included, and optional Care at $677 a year after that. On subscription, the box is $2,260 once and the Retail Counter licences are $1,044 a year, which is $87 a month.
Where store cameras already reach servers you run, the box drops out: the software is $1,260 a year per three-camera store, $105 a month, in a deployment of several stores.
Then one lever, kept small on purpose. Suppose the counts lead to a single change, such as opening a till before the queue builds, and it turns one extra visitor in every hundred into a buyer. Each of those buyers brings in the gross margin on your average basket. On subscription the monthly cost is $87, and the store covers it once its visitors each month reach 100 × monthly cost ÷ margin per basket. To recover the box as well, add its price spread over the months you plan to keep it. Put in your own basket margin, and the entrance count from the first weeks shows whether each store clears that line.
This is an illustration to run with your own figures. ARSA publishes no measured conversion uplift and promises none.
FAQ
AI For Retail FAQ
Do we need a sensor at every door?
No. ARSA uses the cameras already covering your entrances. A new zone needs only a camera that sees it.
How does this compare to the door counters we already have?
Door counters give you a total. This gives you queue length, dwell by zone, and heatmaps from the same cameras, plus the entry count. Most chains run both during transition.
Can we export into our BI stack?
Yes, by scheduled CSV, Excel or PDF export or live REST API.
Does it identify individual or returning shoppers?
Not by default. It counts anonymously: no face templates are created and returning shoppers are not recognised. Repeat-visitor counts are opt-in, with a consent notice; their face data is then sealed on the box and deleted after a set time, 30 days by default. Staff and VIP recognition is opt-in too, for people enrolled with their own consent.
How do we roll out across 200 stores?
If store cameras already reach head office, the rollout is licences on your central servers, with no new hardware in the stores. Where they do not, one AI Box Mini per store, shipped configured, connected in about 15 minutes.
Does store video leave the premises?
No. Counting runs on an AI Box in the store or on servers you own, and only the counts, queue times and dwell figures travel on to head office or your BI stack. ARSA runs no cloud service. In the store it runs as a people counting camera system; centrally, as people counting software.
How do I test people counting on my own store cameras?
Send up to three one-minute clips from an entrance or till camera through the free AI assessment. An ARSA engineer runs the counting model on them and returns an annotated clip, a report that compares the model’s count with a manual count of the same clip, and a 30-minute call, within 3 business days.
Can it count a store with several entrances?
Yes. Each entrance camera gets its own counting line, and the counts roll up to a store total and then across the estate, so you see every door and the whole store. An AI Box Mini handles up to three camera streams, enough for a main and a service entrance; larger stores and supermarkets use the AI Box Server.
Can it alert staff when the queue at the tills gets too long?
Yes. A queue alert fires when a queue zone passes the length you set, so a manager can open another till during the rush. Queue length and wait time are logged per queue zone and per hour, which shows when each store needs more people on the floor.
Talk With Our Consultants
Send us up to three one-minute clips from your own store cameras and an ARSA engineer returns an annotated clip, a report with the numbers and a 30-minute call within 3 business days, free. When you are ready to plan a rollout, a feasibility assessment measures our models against your stores’ lighting, angles and constraints, and its fee is deducted from the project.
Every deployment follows the same steps: a site survey, a check of each camera against our placement standard, installation on an AI Box in the store or on your own servers, tuning on your footage and a handover. Video is processed on site and works without an internet connection, and the clips you send for an assessment are never used to train our models. ARSA ran a queue and service quality analytics proof of concept in Bank Rakyat Indonesia branches.
Related reading: benchmarking foot traffic across store locations, footfall counts for stores with several entrances, reducing queue wait times and occupancy counting for fire code compliance.