People count by eye
A few days of samples. Not every store, every day.
saai count · Trade-area analytics · on-the-ground data
Is it the trade area,
or is it the store?
We count the people who walk past your door and the people who open it, separately. The trade area gets one score and the store gets another.
- Footfall · walked past
- Entry · walked in
- Camera view
The camera looks outside the store, not inside. Stay on the sidewalk and it is footfall. Open the door and it is an entry.
- Footfall
- 1,160
- Entries
- 382
- Capture rate
- 33%
People who walked past the door
People who opened the door
Entries ÷ footfall
No identity is kept. Video is deleted after analysis and only the statistics remain. * Figures are examples.
01 · Limits of trade-area surveys
There have been two ways to measure the front of a store
A 50m grid is the limit
It tells you Gangnam Station is busy. It does not tell you whether your door is busier than the shop beside it.
These are values estimated from cell signals. They count subscribers, and weighting is a correction, not a measurement.
Why it was only ever a sample
Have people measure a 24-hour store around the clock
and it runs about 8.98 million KRW per site each month
One person watches one site, so the cost multiplies by the number of sites. At 24 hours, ten sites run about 77.4 million KRW a month and a hundred sites about 774 million.
saai count only needs more cameras. There are no shifts and no night premiums.
- Layer 1Base wage · 10,320 × 12 hours123,840 KRW
- Layer 2Travel time · + 10,320 × 1 hour134,160 KRW
- Layer 38 night hours · + 10,320 × 1.5 × 8 hours299,280 KRW
12-hour store · month
about 4.02 million KRW
24-hour convenience store · month
about 8.98 million KRW
Assumptions · 10,320 KRW an hour (2026 minimum wage, Ministry of Employment and Labor) · 50% night premium (Labor Standards Act Article 56) · 1 hour of travel a day, round trip
The 4 major insurances and management costs are left out, so this stays below the real figure.
saai count measurement
From the footfall outside,
to the customers inside.
One camera measures footfall, entries, and capture rate every day.
02 · Two axes
The answer splits on two numbers
Footfall is the trade area’s score, and capture rate is the store’s score.
Put the two together and every store lands in one of four places. Which place it lands in is the prescription.
Low footfall · high capture
Trade-area problem
The store did its part.
High footfall · high capture
Model store
Use it as the opening benchmark.
Low footfall · low capture
Exit or renew
When both axes are low, look at the site again.
High footfall · low capture
Store problem
Check the facade, signage, and display.
Neither carrier data nor payment data knows about entries. That is why they cannot build the second axis.
What HQ receives is not a dashboard. It is a table that ranks every store on the same two axes.
Sub-metrics behind the two axes
- Time of day
- Gender
- Age
These explain why the two axes came out the way they did. Figures are examples.
03 · Three horizons
Three horizons, one baseline
Measure operating stores every day and your brand builds a baseline of its own. A new site is read against that baseline.
The three differ only in how long you measure. Every value lands on the same baseline, and the more stores you add the sharper it gets.
Short
A few days is enough
Expo booths · pop-up stores · temporary mall space. You rent it for as long as you rent the spot. Short measurements still land on the same two axes.
Here, no wiring is the first value. If it needs cabling work, this use case does not exist at all.
Validate
A candidate site has no store yet
With no store to measure there is no capture rate. You measure footfall alone and read it against the baseline your existing stores built.
That is what closes the loop. Existing stores have to be measured first for a candidate site to mean anything.
Always-on
Measure daily and a baseline forms
Footfall and capture rate accumulate per store, day by day. The baseline is what your own stores produced, not a category average.
Here, being light is a condition. Without construction or maintenance, a simultaneous multi-store rollout holds.
04 · Install and measure
Three steps on site, three steps in the app
Install · on siteOne outlet is all it takes
01
Mount the camera
Use a suction mount on a wall or glass to frame the street and entrance
02
Plug in the AI box
One indoor outlet; it powers up on its own
03
Connect your phone
Set up from the phone browser. No keyboard or mouse needed
Measure · in the appSet the lines and it starts
04
Set lines
Set the footfall line and the entry line
05
Daily aggregation
Footfall, entries, and capture rate accumulate
06
Report
Review by daypart, gender, and age. Download CSV
No construction, no maintenance. So a rollout to 100 stores does not stretch the schedule by 100 times.
05 · Evidence
Evidence that it counts right, and that it keeps nothing
It counts right
95–96%
Counting accuracy compared against a manual hand count.
- Convenience store 2025 Q1
- 95.8%
- Car showroom 2025 Q3
- 96.6%
- Cultural complex 2025 Q2
- 95.1%
* Counting accuracy vs. our own manual hand count · per-site verification · measurement period noted per item. Results may vary with on-site conditions.
It keeps nothing
Privacy by Design
Anonymization comes first. The SEAL engine
- No video stored
- Deleted after analysis, stats only
- No originals kept
- Source footage is never retained
- No re-identification
- Identity removed, flow only
Read through sense, analyze, and think
One measurement reads three ways
care senses it. insight analyzes it. agent proposes what to do next and connects it to the action.
- saai careSense
What needs attention?
Live occupancy and crowding, with alerts when a threshold is crossed
- saai insightAnalyze
Why did it happen?
Daily, weekly and monthly footfall trends by weekday, hour and year over year
- saai agentThink
What should we do?
Staffing, opening hours and promotion timing proposed from forecast demand
If you need what happens inside
Dwell, gaze, and response at the shelf are read by saai insight. It picks up where count stops at the door.
Where it leads
The same camera signal extends to queue and crowding measurement and media response.
All products