Attention
Webcam gaze mapped onto the shelf you photographed. Camera frames never leave the shopper’s laptop.
Method
What we measure and how
Attention, behaviour and decision at the shelf, not opinions.
Behaviour
Noticed, picked up, put back, carted, kept. Logged during the walk, not recalled afterwards.
Decision
Choice share per version with its margin, and the why from your tasks and a short survey.
Task generator
Your question in, a research protocol out.
Built with the researcher, so every protocol matches what the study needs to answer.
Start from your question
Write the research question and pick the products that matter.
Pick the study type
Packaging, shelf, price, assortment and more. Each type seeds the tasks its question needs.
AI writes the wording
A fixed four stage script: practice, free shopping, directed tasks, questions. The AI writes inside locked slots.
Checked before launch
Every task maps to a measure, and the script is checked for coherence. Edit anything you like.
Evidence
800+ tests in reconstructed worlds.
Our team has collected and analysed data from 800+ consumer tests in reconstructed shopping environments, with global FMCG and retail brands.* Three examples, clients anonymous.
Confectionery
Where should the bar live on the shelf?
Seconds to find the product
Next to wafers5.75 s
Next to count lines7.0 s
Found faster next to wafers, and 50% of shoppers file the bar under a different category than its shelf.
n = 26 shoppers · 2 shelf versions · with a national grocer
Seafood
Which price ladder holds share?
Purchases of the focal brand
Low ladder33
Mid ladder34
High ladder29
Price thresholds per SKU without an in store test. Competitors took 13 to 16 purchases per ladder.
n = 36 shoppers · 3 shelf versions
Personal care
Does the pack shape read as premium?
Seconds of attention per pack, lab eye tracking
Focal pack A28.9 s
Focal pack B27.6 s
Other fourup to 20.2 s
The angular bottle won on attractiveness but not on purchase intent. The unmet need: a pump.
n = 30 women aged 25 to 45 · 6 brands
* Proven at 26 to 36 shoppers per study for planogram, price and pack decisions. TestCart AI shows the margin of every figure and stamps a report decision grade only when its key comparison is settled.
Limits
What we do not claim.
No sales forecasts
We do not predict sales volume.
No blended numbers
Real and simulated shoppers are never merged into one figure.
No lab grade gaze claims
Webcam gaze is never presented as equal to lab eye tracking.
No grade by headcount
Decision grade only when the key comparison is settled.
See the method on your own shelf.
Or open the example report first.