Only the purchase remains
Traffic and gaze in front of the piece never appear anywhere in POS. A number after the sale cannot tell you why it sold or where it lost people.
saai ads insight · Retail media · point-of-purchase ads
How much did
this spot actually get seen?
We count, separately, who walked past the piece, who turned and drew closer, and where a gaze stopped. The value of a spot is scored by response on the floor, not by outcome.
- Traffic · walked past
- Approach · turned closer
- Attention · gaze stopped
Walk straight through the aisle and it is traffic. Turn and draw closer and it is approach. Stop in front of it and your gaze is attention.
- Traffic
- 1,240people · daily
- Approach · 25.6% of traffic
- 318people
- Attention · 44.7% of approach
- 142people · 3.1s avg
- Action · optional signal
- 46pickups
No identification: only gaze and response remain. Video is deleted after analysis. * Figures are examples.
01 Limits
The limit of scoring an ad spot by outcome alone
The spot has no evidence
A near-purchase placement should be priced by the response it actually gets on the floor.
Today, pricing is set by store location and visitor count. Which spot inside the same store got more attention never makes it to the negotiating table.
Same store, same period
ExampleThe cost per attention
differs 3.4× between two spots
With the same visitor count, both spots sell for the same price today. Measure attention and it becomes clear the two spots are different products.
saai ads insight keeps traffic, approach, and attention per spot. Pricing evidence now lives inside the spot, not outside the store.
- Spot ATraffic 1,240 · attention 14211.5% attention rate
- Spot BTraffic 1,190 · attention 403.4% attention rate
- RateSame 600,000 KRW/month · per attention4,225 vs 15,000 KRW
Spot A · per one attention
4,225 KRW
Spot B · per one attention
15,000 KRW
Assumptions · 30-day measurement · Same 600,000 KRW/month placement rate · Attention defined as 1.5s+ gaze
All figures are examples. Real values replace them once measured per placement.
saai ads insight measurement
From who passed by,
to gaze that stayed.
Measure traffic, approach, attention, and action per piece. The next placement has evidence.
02-1 Four stages
Response is not one number. It is four stages
Which stage loses people is the prescription itself.
Low traffic means move the spot. Low approach means check eye level and orientation. Low attention means look at the creative. The same "underperforming" splits three ways.
Traffic
How many walked past the piece, accumulated hour by hour.
Approach
The share of passersby who turned and drew closer.
Attention
Where the gaze stopped, and how long it stayed there.
Action
Picked up or passed by. An optional signal.
38%
1 / 4
* Figures are illustrative.
02-2 Simulation
Same traffic, only the spot changes
Run the measured traffic through two placements and it splits apart: is this a creative problem, or a placement problem?
Same traffic · same creativeExample
12% approach
- Approach rate
- 12%
- Avg. attention
- 1.4s
- Action rate
- 5%
26% approach
- Approach rate
- 26%
- Avg. attention
- 3.1s
- Action rate
- 11%
A · Reading the end-of-aisle wall
Traffic is high but the walk never stops
Traffic alone makes this look like a good spot. Approach stays at 12% and attention cuts off at 1.4 seconds, so a new creative will not hold the eye any longer. People are already facing the direction they are walking.
The simulation runs on measured values. A placement plan run without real measurement is not evidence. Measure traffic, approach, and attention per spot first, then compare the next placement with those values. People confirm it.
03 Evidence
Evidence that it counts right, and evidence that the judgment changes
It counts right
92–95%
Accuracy compared against manually judged approach and attention.
- Traffic & approach 2025 Q2
- 94.6%
- Attention zone 2025 Q2
- 92.3%
- Pickup 2025 Q3
- 90.8%
* Figures are examples. Real accuracy varies with camera angle, lighting, and shelf shape.
The judgment changes
Where the judgment flipped
ExampleThe most expensive spot to sell turned out to be the least seen
It had the highest traffic in the store, but the walk never stopped there. A small shelf next to checkout was pulling three times the attention. Pricing moved from traffic to attention.
Placements measured
Example260 placements · 6 months
Written as category and scale. The brand name is not stated.
Cumulative measurement
Example8.4 million attention records
Cumulative records of traffic, approach, and attention, counted apart.
04 Privacy
Every measurement uses only anonymized spatial data.
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
CV reads the floor while SEAL removes identity.
05 Install and measure
Three steps on site, three steps in the app
Install · on siteOn the CCTV you already use
01
Check the frame
Just check that the piece and the aisle in front of it fit in one frame
02
Plug in the AI box
One outlet next to your existing recorder. No new hardware
03
Connect your phone
Set up from the phone browser. No keyboard or mouse needed
1 / 3
Measure · in the appMark the piece and it starts
04
Mark the piece
Mark signage, shelves, and displays as zones on screen
05
Set thresholds
Set approach distance and attention duration per placement
06
Placement report
Performance per posting, reports by daypart and audience mix, and CSV
1 / 3
The shorter a placement’s rotation, the more it matters to keep results per posting. The last posting’s response becomes the evidence for the next one.
Where it goes
As measurement builds up, the spot becomes a market
01
Targeting
Review exposure by daypart and audience mix. The mix is an anonymous estimate.
02
Objective results
See exposure, attention, and conversion as numbers. Use them in placement pricing negotiations.
03
Bidding
Build toward a market for buying and selling shelf-front placement, like online ads.
Through detect, analyze and act
One measurement reads three ways
care detects the same piece, insight analyzes it, and agent connects it to the action.
- saai careSense
What needs attention?
Live gaze response in front of displays, POP and signage
- saai insightAnalyze
Why did it happen?
Attention per piece analysed (gaze→pick, approach conversion)
- saai agentThink
What should we do?
Golden-zone, placement and exposure priority suggested
1 / 3
If you need the full store route
Dwell, path, and zone transitions are read by saai insight. ads insight picks up where people stop in front of a piece.
Where it leads