Discount, override and loyalty abuse: how AI video can catch it

What discount, override and loyalty abuse is

Every register has tools for changing what a customer pays: markdowns, coupons, price overrides, staff discounts, gift cards and loyalty programmes. Each exists for a good reason, and each can be used to give value away — by a cashier, by a customer, or by both together. Unlike a missed scan, nothing is hidden from the POS: the entry is there, and it looks like an ordinary one. The abuse takes six forms.

Unauthorized markdowns

A price is reduced at the register without the authority to do it: a "damaged" markdown on an item that isn't damaged, an end-of-season price applied before the season ends, or a manual percentage off. The sale is real; the price is not the one the store set.

Stacked coupons

Coupons and promotions are combined where the rules allow only one, applied to items they don't cover, or keyed in when no coupon was presented at all. A cashier can also hold back coupons collected from other customers and apply them later to a sale for themselves or a friend.

Manual price overrides

The item is scanned correctly, then its price is overwritten by hand. Overrides are needed for mispriced labels and price matches, so one extra override is easy to miss. Where overrides need a manager's approval, the approval records a code, not a person: anyone who knows the code can approve their own override.

Staff discounts for non-staff

A staff discount is applied to a sale for a friend, a relative or any customer who isn't entitled to it — sometimes in return for a share of the saving. In the log it looks the same as a staff member buying for themselves.

Gift cards loaded without payment

A gift card is activated and loaded with value, but the money never comes in: the sale is voided after the card is activated, the load is rung through with no tender taken, or a card from the display is loaded and pocketed. The card keeps its balance and can be spent in any store that accepts it.

Loyalty points credited to a staff member's own card

A customer who doesn't present a loyalty card earns no points — unless the cashier scans their own card, or a relative's, on that customer's sale. Over many sales, those points add up to discounts and free items the programme was never meant to pay for.

This guide covers the price tools themselves. Under-rings, items passed around the scanner and the wider pattern of a cashier favouring friends are covered in our guide to sweethearting and missed scans. Refunds and returns belong with return fraud and void and refund fraud.

Why traditional methods miss discount, override and loyalty abuse

Discounts, overrides, coupons, gift cards and loyalty points are part of normal selling. In the transaction log an honest entry and an abusive one carry the same fields: the type of entry, the amount, the register, the time, the employee ID and, sometimes, a coupon, card or loyalty number.

• POS discount and override reports give a count without context. They show who gives the most discounts or overrides, and how much. They can't show what was on the counter, whether a coupon was handed over, or who was on the other side of the register.
• The drawer balances. The customer pays the reduced price, so the cash count matches the POS exactly. When the sale is voided after the card is activated, the drawer shows no gap either.
• Approval codes don't prove presence. A manager's code on an override says the code was entered, not who entered it or whether the manager was at the register.
• Manual CCTV review doesn't scale. Checking one discount means finding the right camera and the right minute. Across every register and every store, footage is pulled only after a loss is already suspected — if at all.
• Classic object detection sees objects, not actions. A detector can say there is a person at the register and an item on the counter. It can't relate what happens at the counter to the discount that was just keyed in — it doesn't know what the POS recorded.

In the log, an abused discount is just a discount. The difference is in what the camera at that register shows when the entry was made: what was on the counter, what was handed across, and who was there.

How vision-language models make it solvable

Primarch builds on and enhances vision-language models (VLMs): models that read video and language together. Instead of detecting objects frame by frame, the model reads the scene at the register the way an experienced loss-prevention auditor would — at every register, every hour.

• It understands the action, not just the objects. A VLM reads a window of time, not a single frame, so the moments before and after a discount, coupon or override entry are read together with the entry itself.
• It joins the video moment with the POS line. Discount, coupon and price-override keys are matched with the motion on camera at that register in the same second. The item on camera is compared with the product on the POS line, so an expensive item rung up as a cheaper one shows as a mismatch.
• It writes the finding in plain language, with a clip. The result is a sentence a manager can read — what was keyed in, at which register, and what the camera shows at that moment — with a time-stamped evidence clip and the matching transaction lines.
• It chains repeated events into patterns. Events are linked by store, register, cashier and hour into 30-day patterns: single events are alerted for review, and the pattern shows what repeats — the same cashier's overrides at the same hours, or staff discounts at the same register.

It matters just as much to say what a check at the register does not show. A camera can't tell whether a buyer is entitled to a staff discount, whether a markdown was authorized, or what a product's price should be; those come from the POS, the price file and the store's own rules. The check shows what happened at the counter when the entry was made, so a manager can judge it against those rules. Gift cards and loyalty points used online or in another channel are outside what a camera at the register can see.

What to look for in a solution

Whichever vendor you talk to, these are fair questions to ask about a system meant to catch discount, override and loyalty abuse:

• Works on your existing cameras. Can it connect to the IP cameras already above your registers, or does it need new hardware?
• Integrates with your POS transaction log. Discounts, coupons, overrides, gift-card loads and loyalty entries live in the log; a camera-only system can't tell which moment to look at. Ask which POS exports it can read, which of these entry types it uses, and how the integration is validated.
• Produces evidence usable in HR processes. A time-stamped clip together with the matching transaction lines, so a finding can be reviewed and discussed fairly — not a bare alert.
• Respects privacy. Is there an on-premise option where footage never leaves the site? Is anonymization applied, is analysis limited to the checkout zone, and does it meet GDPR and other data-protection requirements?
• Can be piloted on your own footage. Your registers, camera angles, promotions and POS — not a demo video.
• Handles false alarms openly. Discounts are a normal part of selling. Ask how findings are reviewed by a person against the store's discount rules, and how single events are separated from repeating patterns.
• Gets alerts to the right role. An instant alert to the store manager, a digest for loss prevention, a dashboard for area and head-office teams.

Example use cases

The walk-throughs below are illustrative examples of how a finding is built. They are not customer cases and contain no performance figures.

Example 1: a price override on a boutique watch

1. 01

   Moment

   A luxury boutique register, 14:20 on a weekday.

2. 02

   What the camera shows

   A watch is placed on the counter, scanned, wrapped and handed to a customer. The cashier keys an entry at the terminal before the customer pays.

3. 03

   What the POS log shows

   One line for the watch, followed by a manual price override to a lower price. No markdown for this item is on record.

4. 04

   The finding + evidence clip

   "At 14:20 a manual price override was keyed on a watch at register 1. The clip shows the watch scanned, wrapped and handed over." The clip and the transaction lines are attached, so the manager can check the override against the store's price rules.

5. 05

   Who is alerted

   The store manager, instantly; loss prevention in the weekly digest.

6. 06

   What the 30-day pattern shows

   Whether overrides repeat on the same cashier's shifts, at the same register, at the same hours — or stay a one-off, such as a price match.

Example 2: stacked coupons at a bookshop

1. 01

   Moment

   A bookshop register, 16:10 on a Saturday.

2. 02

   What the camera shows

   A customer puts four books on the counter; they are scanned and bagged. The cashier keys several entries at the terminal before closing the sale.

3. 03

   What the POS log shows

   Four book lines, then three coupon lines on the same sale.

4. 04

   The finding + evidence clip

   "At 16:10 three coupons were applied to one sale at register 2. The clip shows the counter at the moment each coupon was keyed in." The clip and the coupon lines are attached, so the manager can see what was handed over and check the coupons against the promotion rules.

5. 05

   Who is alerted

   The store manager, in the daily digest.

6. 06

   What the 30-day pattern shows

   Asked in plain language — "show flagged multi-coupon sales at register 2 in the last 30 days" — the event memory returns the list with clips; repeats with the same cashier at the same hours stand out.

Example 3: a gift card loaded and the sale voided

1. 01

   Moment

   A fashion store register, 19:45, shortly before closing.

2. 02

   What the camera shows

   No customer is at the counter. The cashier takes a gift card from the display, swipes it at the terminal and puts it in a pocket. No cash or card payment is taken.

3. 03

   What the POS log shows

   A gift-card activation and load at 19:45, followed a minute later by a void of the sale.

4. 04

   The finding + evidence clip

   "At 19:45 a gift card was loaded at register 3 with no customer at the counter and no payment taken; the sale was voided a minute later." The clip and the transaction lines are attached.

5. 05

   Who is alerted

   The store manager and loss prevention, instantly.

6. 06

   What the 30-day pattern shows

   Whether voids at the same register or on the same cashier's shifts repeat over the 30 days.

Example 4: loyalty points on a staff member's card

1. 01

   Moment

   A souvenir shop register, across a week of afternoon shifts.

2. 02

   What the camera shows

   On several sales, the customer pays and leaves without presenting a loyalty card; each time, the cashier scans a card taken from behind the counter.

3. 03

   What the POS log shows

   Each of these sales carries a loyalty card number, so points were credited on every one.

4. 04

   The finding + evidence clip

   "Loyalty points were credited at register 1 to a card the cashier scanned, not one presented by the customer." Each sale's clip and transaction line are attached.

5. 05

   Who is alerted

   Loss prevention, in the weekly digest.

6. 06

   What the 30-day pattern shows

   Asked in plain language — "show sales credited to this loyalty card in the last 30 days" — the event memory returns the list with clips, by register and hour.

How Primarch's Retail Fraud Expert helps

This section is about our product. The Retail Fraud Expert is the Primarch expert module for the checkout. It recognizes actions at the register and cross-checks each one against the POS record in the same second. Discount, coupon and price-override abuse — including a staff discount used for others — and gift-card and loyalty-point fraud are among the cases it is used for. For these it:

• Matches discount-key anomalies with the motion on camera — discounts, coupons and price overrides are each checked against what the camera at that register shows in the same second.
• Compares the item on camera with the product on the POS line, so an item rung up as a cheaper product shows as a mismatch.
• Chains events by store, register, cashier and hour into 30-day patterns: single events are alerted for review, and the 30-day patterns show what repeats, so a single event isn't treated as a verdict.
• Recognizes whether a coupon was handed over when a coupon line is entered.
• Recognizes a gift card being activated or loaded with no payment taken.
• Recognizes whether a loyalty card was presented by the customer or scanned by the cashier, and chains sales by loyalty card number.
• Produces an evidence file for each finding: a time-stamped clip plus the POS record, usable in HR processes.

It works at the registers of any store with a register — fashion and luxury boutiques, souvenir shops, bookshops — and at restaurant and café counters, wherever a camera covers the register. For friends served at the counter and items passed around the scanner, see sweethearting and missed scans; for a sector view, see retail store loss prevention.

Putting it to work

1. 01

   Connect

   Connect your existing RTSP cameras and the POS transaction log export; no new cameras or special hardware. It is designed to work with any common POS system that can export transaction logs; the integration is validated together during the pilot.

2. 02

   Pilot on your own footage

   Run it on your own registers, camera angles, promotions and POS data, and review the findings together.

3. 03

   Roll out

   Extend to more registers and stores, with central monitoring across the chain.

• Alerts, digests and dashboards. Instant notifications, periodic digests and role-based dashboards, so each finding reaches the person who acts on it.
• Ask the past in plain language. Findings are written into a queryable event memory: "Which price overrides were flagged at register 1 last week?"
• Evidence for HR. Every finding carries its time-stamped clip and the matching transaction line.
• Privacy by design. With on-premise deployment, footage never leaves the facility. Analysis focuses on the checkout zone with privacy-compliant anonymization; the goal is action–transaction consistency, not tracking people.

Industry figures

There is no reliable public figure for discount, override or loyalty abuse on its own. The ACFE's estimate covers occupational fraud as a whole and is not limited to the checkout.

A discount that looks like any other leaves no gap in the drawer. Checking each discount, coupon and override entry against the footage turns it into a specific moment that can be reviewed; repeats at the same register and hour stand out as a pattern.