In-store camera analytics: 10 privacy questions
Ten questions to ask any in-store camera analytics vendor before you install: where images are processed, what is kept, what reaches the cloud.
- Home
- Privacy
Anonymous in-store analytics: images stay in the store, faces are never recognised
Where GoTrack processes camera images, what it keeps and what reaches the cloud. This page describes technical behaviour; it is not legal advice.
GoTrack analyses camera images on a device inside the store and does not store them by default. It uses no face recognition, creates no biometric data and does not recognise shoppers between visits. The cloud receives product events — which product, which zone, when, how long it was held — never video.
Four principles
Processing on siteAnalysis happens in the store device's memory. Recognition and screens keep working without internet.
Nothing stored by defaultFrames are discarded after analysis. Debug recording is off unless someone switches it on, and then stays on the device and is deleted after 30 days.
No identityNo faces, no face templates, no age or gender estimation today, no recognising a returning shopper.
Minimal cloud dataProduct events and device health reach the cloud. Images and video do not.
What reaches the cloud
| Sent to the cloud | Never sent |
|---|---|
| Product events: product, variant, zone, camera, time, hold time, put-back | Camera images and video |
| Device health: software version, camera status, disk, error counts | Faces, face templates, biometric data |
| Your product catalogue and the product photos you upload | Names, phone numbers, loyalty cards or any shopper identity |
The details
Images reach the cloud in two cases, and an authorised user starts both. First: if the user chooses to add a product photo taken by a store camera, that image is saved to the product catalogue. Second: if the user asks GoTrack to draw shelf slots automatically during setup, a small close-up of each product is sent to the dashboard for approval; no view of the store is sent, a close-up that overlaps a person is never sent, and the close-ups are deleted once the slots are saved. Neither happens automatically.
Short-lived tracking numbers. To tell that a product is being held, GoTrack looks at where people and hands are in each frame. The numbers used to follow them between frames belong to one camera, last seconds and identify no one.
Staff areas. GoTrack has no pixel-level privacy masking. Analysis is limited to the shelf zones defined for each camera; today the way to keep staff areas out of view is to aim the camera at the shelf and the shopping area only.
Audience insights and your own assessment
We are building audience insights: which age ranges and genders pick up each product, as aggregate reports. They are not yet running in stores. Before they are released, this page will describe exactly what they process and what they keep.
Even without storing images, camera analytics processes images in real time. The retailer, as data controller, should review notices and records with its own legal team; this page can be used as input. Ask GOTONOM in writing where cloud records are hosted and how long they are kept.
For any vendor, GoTrack included, use the 10 privacy questions to ask before you install.
Privacy questions
Does GoTrack use face recognition?
No. There is no face recognition, no face template, no biometric matching, and no shopper is recognised across visits.
Does GoTrack record camera footage?
Not by default: each analysed frame is discarded. If debug recording is switched on, the files stay on the store device and are deleted after 30 days by default. The only image that reaches the product catalogue is one an authorised user deliberately adds as a product photo.
Does GoTrack work without processing personal data?
GoTrack does not store images, identify anyone or send personal data to the cloud, but it does process camera images in real time on the store device. How that is classified under KVKK or GDPR, and the notices it needs, is for the retailer's legal team to assess as data controller.
Do shoppers need to be informed?
Yes. The retailer, as data controller, informs shoppers, and the notices should describe the analytics, not only security cameras. GOTONOM provides the technical description your legal team needs.
Can GoTrack mask staff areas and analyse only the shopping area?
There is no pixel-level privacy masking. Analysis is limited to the shelf zones defined for each camera. Today, keeping staff areas out of view means aiming the camera at the shelf and the shopping area.
Where is the data kept?
Camera images are processed on the store device and, by default, never leave the store. For the hosting region and retention of cloud records, ask GOTONOM in writing before purchase.
Can GoTrack show the age and gender of the shoppers who engage with a product?
We are building it now. Audience insights will show which age ranges and genders pick up, hold and put back each product, as aggregate reports, never as individual profiles. It is not yet running in stores, so GoTrack does not estimate age or gender today; tell us if you want it in your pilot.
Further reading
Bring your data protection officer.
We are happy to walk your legal and security teams through GoTrack in detail.
or write to info@gotonom.com