Self-Checkout Is a Failed Experiment

This article was featured in One Story to Read Today, a newsletter in which our editors recommend a single must-read from The Atlantic, Monday through Friday. Sign up for it here.

When self-checkout kiosks began to pop up in American grocery stores, the sales pitch to shoppers was impressive: Scan your stuff, plunk it in a bag, and you’re done. Long checkout lines would disappear. Waits would dwindle. Small talk with cashiers would be a thing of the past. Need help? Store associates, freed from the drudgery of scanning barcodes, would be close at hand to answer your questions.

You know how this process actually goes by now: You still have to wait in line. The checkout kiosks bleat and flash when you fail to set a purchase down in the right spot. Scanning those items is sometimes a crapshoot—wave a barcode too vigorously in front of an uncooperative machine, and suddenly you’ve scanned it two or three times. Then you need to locate the usually lone employee charged with supervising all of the finicky kiosks, who will radiate exasperation at you while scanning her ID badge and tapping the kiosk’s touch screen from pure muscle memory. If you want to buy something that even might carry some kind of arbitrary purchase restriction—not just obvious things such as alcohol, but also products as seemingly innocuous as a generic antihistamine—well, maybe don’t do that.

All is not rosy in the world of self-checkout, and some companies seem to realize it. Walmart has removed the kiosks entirely from a handful of stores, and is redesigning others to involve more employee help. Costco is stationing more staffers in its self-checkout areas. ShopRite is adding cashiers back into stores where it had trialed a self-checkout-only model, citing customer backlash. None of this is an indication that self-checkout is over, exactly. But several decades in, the kiosks as Americans have long known them are beginning to look like a failure.

Before self-checkout’s grand promise could be sold to the general public, it had to be sold to retailers. Third-party firms introduced the kiosks starting in the 1980s, but they didn’t take off at first. In 2001, when the machines were finally winning over major retailers in masse, K-Mart was frank about its motivations for adopting them: Kiosks would cut wait times and allow the company to hire fewer clerks. Self-checkout is expensive to install—the average four-kiosk setup runs around $125,000, and large stores can have 10 or more kiosks apiece. But write one big check up front, the logic goes, and that investment eventually pays off. Human employees get sick, ask for raises, want things. Computerized kiosks always show up for work, and customers do the job of cashiers for free.

Except, as the journalist Nathaniel Meyersohn wrote for CNN last year, most of this theory hasn’t exactly panned out. The widespread introduction of self-checkout kiosks did enable shoestring staffing inside many stores, but it created plenty of other expenses too. Self-checkout machines might always be at work, but, on any given day, lots of them aren’t actually working. The technology tends to be buggy and unreliable, and the machines’ maintenance requires a lot of expensive IT workers. Much of the blame for that can be placed on the systems themselves. During the years I spent processing purchases at big-box and chain retailers in the 2000s, every point-of-sale system I used felt more intuitive and less error-prone than the ones I’m now regularly tasked with navigating as a paying member of the public.

There’s little evidence that self-checkout is reliably faster than the old cashier system, and that feel of convenience has always been largely a trick of perception. “Trained cashiers can scan and bag goods faster than even the most aggressive or enthusiastic shopper,” conceded a 2002 New York Times story about the machines’ growing popularity. “But actual checkout speed tells only part of the story. Self-checkout has a psychological effect: as long as the shopper is taking an active part, it seems to go faster.”

Perhaps an even larger issue than the problems that self-checkout directly creates is the set of behaviors its presence can enable—from executives, from employees, and from customers. Retail executives, looking for any available corner to cut in order to juice short-term profitability, took self-checkout’s proliferation as a license to trim store staffing to the bone. Many stores are now messier, their shelves go unstocked for longer, and customers have a harder time finding the products they’re looking for or employees to answer their questions. Retail jobs, which have long been low-paying, precarious, and unpleasant, are now even worse. Forget staffing up to the levels required for a good customer experience; in some instances, stores have reportedly had problems staffing their floors to a minimum level of employee safety. Self-checkout is far from solely responsible for this doom loop, but it has contributed to the way certain stores have rotted from the inside.

And then, of course, there’s theft. Not only does the act of bagging up your own stuff create new opportunities to make it out the door without paying for everything, but understaffed stores also enable theft overall. The most reliable way to deter shoplifting is to make thieves think they’re going to get caught, but when even customers who want to pay for something struggle to flag down an employee, the store has already forfeited that battle entirely. Theft, as well as the losses from the unscanned and mis-scanned items that poorly designed kiosks create, is a trade-off of which retailers are well aware. (Companies contacted for this story, including Walmart, CVS, Costco, and Kroger, either did not respond or declined to comment on self-checkout’s relationship to theft.)

But, of course, stores don’t really mention self-checkout when they panic about retail theft. Instead of hiring enough people to run a functional store, companies push the task of deterrence onto customers who are already doing the scanning and bagging work of a cashier. By generating widespread fear of a retail-crime spike, stores that rely on self-checkout can deputize the public and put pressure on law enforcement to become more visibly involved: People become more suspicious of the shoppers next to them, local police feel obligated to change patrols to ensure that stores are safe, and governments pour additional public money into theft prevention. Not that there is a lot of data to suggest that this oft-discussed recent spike in retail crime is even real. In an earnings call earlier this year, for example, the CFO of Walgreens admitted that his company had perhaps “cried too much” about theft, and that its product losses had actually declined in the previous quarter. (Walgreens declined to comment for this story.)

Some retailers seem to realize that they may have overcorrected in their swing toward unmanned checkout, and are adding more human labor back into the checkout process on a trial or permanent basis. Still, self-checkout kiosks as they are currently constituted are likely to persist for years to come, because the machines were so expensive to install in the first place. Kroger is forging ahead by testing a store format that relies entirely on self-checkout, but the grocer has said that the move isn’t expected to result in any labor savings over its standard checkout structure, which suggests that there won’t be fewer cashiers. More and more, self-checkout has begun to refer to the style of kiosk instead of to the expectation that everyone who uses it will conduct the transaction alone. There is now less “self” in self-checkout.

The kiosk technology, as it stands, just isn’t good enough to support the level of autonomy from the vagaries of paid human labor that retailers wish their stores could have. Within the industry, there are great hopes for its eventual improvement; Amazon, for example, continues to tinker with its Amazon Go store format, which requires you to check in to the store from the company’s app but lets you leave with no checkout at all. Perhaps that is the future, but for now, a familiar limitation of many grand tech-industry promises endures: At the bottom of all the supposed convenience, you do actually just need a lot of people to operate a store.