AisleWatch AI

AisleWatch AI / Product lines / Retail loss prevention

Retail loss prevention AI that catches theft before the stock count.

Most losses only show up at the stock count, weeks after the goods have gone. AisleWatch AI shows your team theft and till fraud as it happens in every store, with a short clip for each alert.

Alert stream, all storesToday, newest firstLive
  1. Store 014, aisle 4Item possibly hidden, sent for review◐ Acknowledged
  2. Store 017, till 3Refund with no customer at the till✓ Closed
  3. Store 014, till 2Drawer open with no sale◐ Acknowledged
  4. Store 045, aisle 6Several items swept from a shelf, sent for review✓ Closed
  5. Store 031, rear doorPerson in the store before opening✓ Closed
Simulation.

Retail theft and shrink

Loss prevention teams usually review footage after a loss is found, one store at a time. Patterns across stores stay hidden in separate recorders.

  • Losses are found at the stock count.
  • Footage is pulled store by store.
  • Nobody records which alerts were real.

65% of US retail shrink came from external and internal theft (financial year 2022). Source: NRF National Retail Security Survey 2023 (third-party public figure)

Loss prevention by store risk

Shoplifting detection

Items hidden or swept off high-loss shelves go to staff for review.

Cashier and POS video

No-sale opens, refunds and voids are checked against the till camera.

Jewellery store security

Staff hear when a case is open with nobody behind the counter.

AI store security cameras

Theft, till and after-hours alerts suit convenience stores and small chains.

Fuel stations

Fills are matched to payments, and cash is watched at shift change.

Case records

Clips and times are kept for each case you report to the police or insurers.

Retail loss prevention alerts, checked by a person

The store team gets each alert with the clip. Loss prevention sees every case across stores.

AisleWatch flags behaviour for review. A store team member checks every alert before anyone speaks to a shopper, and false alerts are reported each week of the pilot.

Shoplifting detection camera still: a shopper with a basket beside the high-loss bay marked for review, face blurredSimulation.

Real-time alert16:41:12

A shopper may have hidden an item at the high-loss bay.

Aisle 4, Store 014

Sent to
Store team, Store 014
Channels
Microsoft Teams, SMS, email and the dashboard
  1. 16:41Store team on the floor, by Teams
  2. 16:43Store manager, if nobody responds
  3. 16:46Area manager, if still open

Loss prevention across the estate

Loss prevention sees every store in one week, with every alert the store teams confirmed or marked false. The figures below are a simulation with sample data.

  • 95 alerts reviewed across 5 stores.
  • 29 confirmed by store teams.
  • 66 marked false, in the weekly false alert count.
  • 4 cases open now.
Loss prevention, all stores, week 46Alerts by type, with alerts confirmed by the store team and cases open now
Hidden itemsTill exceptionsShelf sweepsAfter hours
0102030Store 014Store 014: 27 alerts, 8 confirmed, 2 open27 alerts, 8 confirmed, 2 openStore 017Store 017: 26 alerts, 10 confirmed, 1 open26 alerts, 10 confirmed, 1 openStore 045Store 045: 20 alerts, 6 confirmed, 1 open20 alerts, 6 confirmed, 1 openStore 022Store 022: 12 alerts, 2 confirmed, 0 open12 alerts, 2 confirmed, 0 openStore 031Store 031: 10 alerts, 3 confirmed, 0 open10 alerts, 3 confirmed, 0 open
Store 01427 alerts, 8 confirmed, 2 openStore 014: 27 alerts, 8 confirmed, 2 openStore 014: 27 alerts, 8 confirmed, 2 openStore 014: 27 alerts, 8 confirmed, 2 openStore 014: 27 alerts, 8 confirmed, 2 openStore 01726 alerts, 10 confirmed, 1 openStore 017: 26 alerts, 10 confirmed, 1 openStore 017: 26 alerts, 10 confirmed, 1 openStore 017: 26 alerts, 10 confirmed, 1 openStore 017: 26 alerts, 10 confirmed, 1 openStore 04520 alerts, 6 confirmed, 1 openStore 045: 20 alerts, 6 confirmed, 1 openStore 045: 20 alerts, 6 confirmed, 1 openStore 045: 20 alerts, 6 confirmed, 1 openStore 045: 20 alerts, 6 confirmed, 1 openStore 02212 alerts, 2 confirmed, 0 openStore 022: 12 alerts, 2 confirmed, 0 openStore 022: 12 alerts, 2 confirmed, 0 openStore 022: 12 alerts, 2 confirmed, 0 openStore 03110 alerts, 3 confirmed, 0 openStore 031: 10 alerts, 3 confirmed, 0 openStore 031: 10 alerts, 3 confirmed, 0 openStore 031: 10 alerts, 3 confirmed, 0 open
Simulation.

One platform for loss prevention

Every check runs on the store cameras and reports into one case record across stores. Till and self-checkout checks can link to the POS by API, scoped for each retailer.

Loss prevention capabilities

  • High-value cases and after-hours entry● Live alert with the clip
  • Concealment and shelf sweep alerts● Live alert at the shelf
  • Till exceptions matched to POS● Matched to the POS
  • Self-checkout checks● Flagged for staff review
  • Cases linked across stores● One record across the estate
  • Repeat offender watchlist⊘ Optional, switched on by the retailer, off in the EU and Australia

Questions about retail loss prevention

Send the AisleWatch team another question

What is retail loss prevention AI?

Retail loss prevention AI watches store cameras for theft, till fraud and break-ins, and it sends staff the clip. AisleWatch also keeps a record of every case.

How does video help retail loss prevention?

AisleWatch flags possible theft and till problems while they happen, so the store team can act before the loss is final.

Does AisleWatch accuse shoppers?

AisleWatch never accuses a shopper of theft. It only flags behaviour for a person to check.

What does a loss prevention pilot measure?

The pilot reports alerts checked, false alerts each week, cases recorded and till problems confirmed. The measures are agreed before the 8 weeks start.

Can AisleWatch cover self-checkouts?

AisleWatch watches the self-checkout bank and flags moments for staff to review, such as an item bagged with no scan. A link to the self-checkout system by API is scoped for each retailer.

See a sample store report from one day of your own footage.

AisleWatch returns the alerts one day of your footage would have raised. The footage is covered by an NDA and deleted after review.