Cashier fraud detection and prevention, now made easy with A.I technology

by Anh H. Nguyen

Internal theft has been plaguing retailers since time immemorial

As dismaying as it may be, the truth remains that employee theft is prevalent, persistent, and almost impossible to get rid of completely. Broadly defined, employee theft is any stealing, misuse, or abuse of a company's assets, whether tangible like cash and merchandise, or intangible like time and information. The numbers are distressing:

- In the US, the retail industry wastes a collective 46.8$ billion each year due to employee dishonesty.

- On average, one in five dollars is lost to frauds.

- Surprisingly, 55% of perpetrators are in managing positions.

- 43% of workers admit to having stolen from their companies at least once.

- A global study on retail theft found out that employee larceny accounts for 4 times the amount of losses incurred by shoplifters. A more recent survey put it at an even higher ratio.

Although similar surveys have not yet been conducted in Vietnam, the figures in our country probably do not fare any better. To add insult to injury, no matter how many resources a company may pour into staff ethics training, the transient nature of retail work means all those efforts often end up futile. The retail turnover rate currently sits at 60%, four times higher than the average for all industries. Not only that, the 10-10-80 rule dictates that 10% of employees will steal no matter what, the other 10% will never steal, and the rest may or may not steal due to circumstances that may be entirely out of the owners' control.

There are many ways for unethical employees to circumvent the POS system and commit theft

The employee theft playbook may be elaborate and long, but fraudulent activities most often happen around the cash register for the obvious reason: that is where most of the actions take place. It is not feasible to bring every types of unsavory practices into the scope of this article, so for the sake of brevity, this is a list of the most common ones:

- Plain cash theft: Cash is king, easy to steal, and highly liquid. A cashier may pocket a customer's payment (or a portion of it) by not entering it into the system, entering only part of it, creating a fake refund, or voiding the transaction. More audacious employees would skim cash directly from the drawer or the petty cash tin, then falsify the records to cover their tracks.

- Promotion abuse: From time to time, some retailers run marketing campaigns such as "Buy one get one free," seasonal sales, or small gifts and vouchers attached to certain purchases. Sneaky cashiers may choose not to mention the discount to customers and swipe the difference. They may also combine every two single transactions to appropriate the free items, or just "forget" to give customers the free gifts and sell them themselves elsewhere for profit.

- Goods stealing: This may be as rudimentary as store employees hiding products in their coats or the trash bags, or "sweethearting" - the practice of letting family members and friends take what they want without paying.

How do retailers combat these unscrupulous behaviors? Many retailers try to cope by offering customers rewards in exchange for reporting no-receipt transactions. Some resort to raising wages in the hope it might lessen the incentive to steal, but not many can afford to do so. Some implement heavier inventory controls. Some assign a surveillance division or hire outside bookkeeping services to monitor transactions, both of which beget more costs. Some simply resign to the fact that putting up with employee theft is an inevitable part of doing business.

A new generation of powerful, AI-backed, anti-internal theft tools have finally arrived

That is no longer the case. Almost every store these days is equipped with CCTV cameras for security reasons, and they are retailers' greatest helping hand. Retailers could rest easy: there is no need to posit a special team dedicated to catching frauds in front of the computer screen all day. Breakthroughs in machine learning have given rise to advanced alert systems that could be perfectly customized to the retailers' needs. After being trained to recognize patterns of dubious behaviors, especially when used in tandem with POS data, the software could automatically detect and flag deceits, then submit them to higher-ups in the form of video compilation.

A high-end clothing chain of 80 stores, for example, would encounter an average fraud rate of 0.75%. With its 40000 transactions per month and the usual bill valuing about 100$ US, the chain suffers a 30000$ US loss annually. After installing a basic AI cashier auditing system for 100$ US/store/month, they could save up to 264000$ US per year. A relevant case study could be found here.

Besides helping the retailers' bottom line, there are many other benefits to using AI technology to minimize retail frauds

Example of a suspicious activity caught on camera

First of all, depending on specific circumstances, business owners could decide to review the footage evidence on a daily, weekly, or monthly basis, reducing the workload and saving time and cost of operation. Besides, by substituting unrelenting softwares for tired human eyes, this alert system is less susceptible to biases and errors.

Secondly, this system could aid retailers in cultivating a much healthier company culture. A Harvard Business Review found that toxic attitude is contagious just like the common cold: 37% of people tend to mimic the bad behaviors of those they work with, much like how the whole barrel spoilt by a few rotten apples. Thanks to the streamlined nature of this system, misconducts could be rectified immediately and the employees at fault swiftly disciplined, effectively halting the spread of wrongdoings.

Thirdly, similar to the Robert Frost's quote, "Good fences make good neighbor," good surveillance systems also make honest employees. Those hell bent on committing acts of frauds would be identified and fired quickly, while those thinking of mischief would be deterred from following through.

Last but not least, the all-seeing camera footage makes the process of reviewing much more thorough and omniscient. From the machinations of their shady employees, retailers could deduce and learn to patch the breaches in their security system, hence minimizing the perils of internal theft with the ultimate goal to eliminate it all together.

Oftentimes, AI technology may sound a bit "out there", giving the impression that it still solely belongs to the future. In reality, like its usage in preventing retail theft, AI applications are real, practical, effective, and increasingly commonplace. The best time to invest in AI solutions is yesterday, the second best time is right now. For more information or to request a product demo, visit:


© Copyright 2020. All rights reserved