How Do You Reduce Employee Theft at the Register?

By awareretailgroup.com Editorial· Edited by Tom Walsh· Published 2026-10-16

Reduce register theft by putting controls on the three transactions where a cashier can move goods or money without a matching sale: voids and no-sales, discount overrides, and no-receipt refunds. Then review the exceptions those controls flag on a fixed schedule, so a pattern tied to one cashier login gets caught while it is small. Pilot each change in one store before rolling it out chainwide.

What NRF’s internal theft numbers cover, and why they’re dated

The National Retail Federation’s 2023 National Retail Security Survey, run with the Loss Prevention Research Council, covers FY2022 loss. It put internal theft at 29% of shrink loss and process and control failures at 27% (NRF survey). Treat those figures as dated: after more than three decades of annual shrink research, NRF did not publish a shrink report in 2024 and pointed to the 2023 survey as the latest available statistics (Retail Dive). The same survey reported merchandise theft, refund fraud, cash and deposit theft, and sweethearting as the top four internal theft methods and an average internal theft loss of $2,180 per investigation (NRF survey). Map your own voids, discounts, no-receipt refunds and cash handling against those four methods before you add any new register control.

Which register controls are backed by evidence, and which aren’t

The controls below fall into two evidence tiers. A few rest on studies or survey data from a named setting. The rest are reasonable responses to what NRF’s survey says retailers lose, with no cited study testing the control itself. None of the studies cited here was run at a retail register, so each row names where its evidence comes from.

The strongest measured result came from outside retail: when 392 restaurants in 39 states added point-of-sale monitoring software, theft dropped 22 percent and revenue rose about 7 percent (Pierce et al.). Build the no-receipt refund control anyway, because the problem it targets is already named in the survey data, and measure it in a later pilot once the exception review is running. Discount and override limits are the least supported row: the cited research names sweethearting as a top-four method but gives no tested override mechanism, so set who can approve an override, and up to what amount, as local policy and measure it in the pilot.

How do I control no-receipt refunds and how often should I review register exceptions?

NRF’s 2023 survey names refund fraud as one of the top four internal theft methods (NRF survey). The refund controls below follow from that finding. None has been tested in the research cited here, so each is a pilot hypothesis.

Set a refund limit your POS can enforce automatically, so the rule holds on a busy shift without anyone having to remember it. Reviewing exceptions on a fixed schedule is this article’s recommendation, and a daily review is the version it favors. ACFE’s 2026 report suggests fraud can be uncovered up to four times faster with active detection methods than with passive ones (ACFE 2026 report), and 43% of the frauds in its study were detected after a tip was reported (ACFE 2026 report). Both figures cover fraud cases across industries. Start the daily exception review this week, then pilot the automatic refund limit once your POS can enforce it.

Why process changes are a better bet than reminders

A meta-analysis by Mertens and colleagues, covering many behavioral domains and not retail, found that interventions targeting the structure of choice alternatives consistently outperformed interventions that described the alternatives or reinforced intentions, a category that includes reminders (Mertens et al.). A later reanalysis by Maier and colleagues, correcting for publication bias, found evidence against an effect for information and assistance interventions and called the evidence for structural interventions undecided (Maier et al.). Neither analysis gives reminder-type interventions an edge. At the register, a poster asking cashiers to be honest works like a reminder, and a system that will not post a refund without a second login changes the structure of the choice. Put the budget into the login requirement first.

Explain new register controls before you enforce them, then back the supervisors who do

A field experiment at a Dutch local government department, outside retail, tested nudges to raise compliance with a policy requiring employees to wear an identifying lanyard badge, where compliance had been low. The researchers found insufficient support for the nudges working, and they hypothesize that the nudges failed to change attitudes about the policy and led to reactance among some employees (Dewies et al.). Announce a new register control in a five-minute huddle with the reason attached, such as refund fraud or drawer accuracy, before you enforce it.

A forest-products mill, also outside retail, set a goal with its union executive committee to cut theft from about a million dollars a year to zero (Latham). Theft dropped to near zero and stayed inconsequential by company accounting methods for three years (Latham). Latham warns the result likely generalizes only to settings where theft is a way to find excitement at work.

Latham recommends that upper management tell supervisors they will be supported for insisting on and monitoring honest behavior (Latham). Tell your supervisors before the rollout that you will back them when they enforce the new rule.

Pilot new register controls for 60 to 90 days before rolling them out chainwide

Pick one control from the partly supported rows. Exception-based reporting paired with a daily review is the strongest combination in the tables above. Run it in one store for 60 to 90 days against a comparison store that keeps its current process. A sibling store makes a better baseline than last year’s numbers, since seasonality and promotions move exceptions on their own.

• Pick one control (exception reporting or review cadence) to pilot first
• Choose a comparison store with similar size, format and staffing
• Set the review cadence and name who owns it before day one
• Track flagged exceptions, refund dollars and drawer variance in both stores weekly
• Set the go/no-go threshold before the pilot starts

Scale the control if the pilot store meets the go/no-go threshold you set before day one on flagged exceptions, refund dollars and drawer variance, measured against the comparison store. If it does not, the control may have missed the problem. Make the scale-up call by day 90.

What to do this week

• Pull last month’s void, no-sale, discount override and no-receipt refund reports and sort them by cashier login.
• Find out whether your POS can enforce a refund limit automatically; that is the later pilot, after the exception review is running.
• Tell cashiers why the daily exception review exists before it starts.
• Assign one person to review flagged exceptions daily, and name who they escalate to.
• Pick a pilot store and a comparison store, and start a 60- to 90-day pilot of the exception review before changing the policy chainwide.