00:00:00/Check

Shrinkage Heatmap: where the losses cluster

Shrinkage Heatmap is a report check in Svid: point it at your footage and it reads the whole timeline, end to end, then writes a report mapping where and when loss events cluster. Instead of guessing from the stocktake number, you see it as a map: the aisle, the fixture, the hour of day.


What it does

This is a whole-video check, not a live alarm. Run it across a week or a month of footage and it reads the lot, noting each theft-shaped moment with its camera, its place and its time, then plots them over your floor plan and your hours of trading and writes the result up as a document. The stocktake tells you how much walked out; the report shows where and when. The two together are the start of an actual plan, rather than a bigger mirror budget applied at random.

What the pattern buys you

  • Moving the high-theft fixture out of the blind corner it turns out to have been sitting in.
  • Putting the guard's hours where the incidents actually are, which might be Thursday evenings and not Monday mornings.
  • A mirror, or a camera, aimed at the aisle the map keeps lighting up.
  • Evidence for head office that the refit request is based on something better than a feeling.

What you get

A written report with the heatmap in it, arriving on screen and by email, and a JSON webhook for anyone feeding a reporting system. Every dot on the map is a described moment with a link, so the underlying footage is always one click away. It pairs naturally with Theft Watch: that's an alerting check that flags the moment while the person is still in the shop, while this one reads the whole video afterwards and learns the pattern over weeks. Against Footfall & Dwell Time the contrast is simpler still: that maps where people go, this maps where losses happen.

Where it runs

Per camera, and the map gets better the more of the floor the enabled cameras cover. It works on RTSP captures, uploads and recorded exports, so you can build the first map from last month's footage before you change anything on the floor. A fully on-prem appliance mode exists for footage that can't leave the building. One thing the map deliberately is not: it's a map of events and places, never of people. Svid doesn't track people or recognise faces, anywhere, on purpose.

What it costs

Report checks are the cheap end of the Store. Before you buy, Shrinkage Heatmap shows a live monthly price computed from your own usage (cameras, hours a day, watch level) as ‘+$X/mo at your usage’, and report checks are priced from a $3 a month floor. It joins a subscription that is priced from your own usage, minimum $9 a month, with the full derivation on screen at signup; the pricing section shows the working.

Where does the heatmap's data come from?
From your own footage. The check reads the whole video and notes each theft-shaped moment with a place and a time, so the map is your footage aggregated, not an industry benchmark dressed up as one.
Can I build a map from old footage?
Yes. Recorded video goes through the same pipeline as live, so a month of exports gives you a first map before you commit to moving a single fixture.
Does it tell me who is responsible for the losses?
No, and deliberately so: it maps where and when events cluster, never who. Svid has no people tracking, no re-identification and no facial recognition, and that's a design decision, not a roadmap item.

Draw the first map from your archive

Create an account at app.svid.ai, upload a month of exports and run the report. You'll see where the losses cluster before you move a single fixture.

Get started

Related footage