The pitch was simple: remove the cashier, remove the line. Over the past decade, self-checkout machines spread through supermarkets, pharmacies, and big-box stores as the visible face of retail automation. Shoppers were told the machine would do the work and the queue would disappear.

That story is now being revised by the retailers themselves. Across the industry, chains have pulled self-checkout out of stores, capped the number of items allowed through the lanes, and rethought what the technology actually saved. The reason is not nostalgia for human cashiers. It is the data, and the data has been unflattering for years.

What the receipts show

Theft and error rates at self-checkout run higher than at staffed lanes. According to a Business Insider review of retail data published in December 2023, 6.7 percent of self-checkout transactions involved some form of shrink, the industry term for lost inventory. The same reporting found that one in five shoppers admitted accidentally taking an item without paying, and one in seven admitted doing it on purpose.

The pattern is not confined to one country. Since 2022, the British chain Sainsbury's has required self-checkout customers to scan a printed receipt at a gate before they can leave, a system criticized for treating every shopper as a suspect until proven otherwise, as reported by the Retail Gazette and The Independent. In the United States, some Target stores responded to a 120 percent rise in retail theft in 2023 by limiting how many items a customer could take through self-checkout, and a few closed the lanes entirely.

The "cashier-free" store that needed 1,000 people

The clearest example came from Amazon. Its "Just Walk Out" technology, installed in Amazon Go stores and later licensed to other retailers, was promoted as a fully automated system: cameras and sensors tracked what you took, and you walked out without stopping.

In April 2024, that claim came apart. Reporting by Business Standard revealed that the system depended on roughly a thousand workers who manually reviewed transactions, checking footage when the software misread them. Amazon then announced it would replace "Just Walk Out" with Dash Carts in its Amazon Fresh stores.

The timeline matters. Amazon announced the concept in December 2016, opened the first Amazon Go store to the public in January 2018, expanded into the larger Go Grocery format in 2020, and then began closing locations in 2023. What was sold as the end of checkout became a case study in how often the automation still needed a person behind the curtain.

Why it matters

For shoppers, the shift is immediate and physical. Self-checkout moves the scanning, the bagging, and the error correction onto the customer, then adds surveillance to check the result. Missed-scan detection and receipt gates mean the honest customer absorbs the cost of the dishonest few, and shoppers have complained that the machines treat them like thieves.

For workers, the effect runs the other way. Labor reporting and commentators have noted that self-checkout lets retailers cut in-store staffing, which trims entry-level jobs and leaves fewer people to help when a machine jams or a price rings up wrong. The Guardian reported in 2019 that shop closures and self-checkout contributed to the loss of tens of thousands of women's jobs in the United Kingdom.

The counterpoint

Self-checkout is not disappearing. Many shoppers prefer it for small baskets, and retailers still install it because it can shorten waits at peak hours. The technology tends to work better in convenience formats than in full grocery trips, where loose produce and bulk items defeat the scanners and force a staff override anyway. Some chains now reserve staffed lanes for larger baskets and steer members with a phone app toward scan-and-go, a hybrid that keeps at least one human within reach.

What is changing is the marketing. The pitch is shifting from "no humans" to "fewer errors," from full automation to assisted checkout. Retailers are learning that the register was not simply a scanner. It was a person who could notice, and noticing turns out to be the hard part to automate.