SEATTLE—I entered Amazon Go, the company’s checkout-less convenience store in Seattle, at 10:23 a.m. Monday and spent precisely 11 minutes and 59 seconds browsing before I walked out with a sandwich and a yogurt cup. This information was available because the moment I scanned a personalized QR code at the store’s subway-style gate, myriad cameras on the ceiling started tracking me. Every time I picked up an item, it was added to my virtual cart; when I placed the item back on the shelf, it was removed. A couple minutes after I walked out, Amazon charged $6.61 to the credit card linked to my account. The receipt included the time-spent-shopping tidbit, presumably to impress customers with how little time it took to buy lunch.
Just as striking as the shopping experience was the scene outside, where there’s evidence that the company famously seeking a second home city is still gobbling up its first. South of the store was territory the company has already colonized—gleaming office towers where 40,000 employees work, expensive apartments where many of them live, glassy domes filled with rare tropical plants. Across the street, four cranes sprouted from a hole in the ground that will eventually anchor another office tower. Construction noise was ceaseless. In front of the store, two masked protesters held a banner that proclaimed, “YOUR FUTURE IS TOTAL SHIT.”
Given what’s already known about automation, labor economics, and Amazon founder Jeff Bezos’s appetite for expansion, there’s little reason to believe the company won’t scale up the innovations tucked into its new convenience store. By further easing the buying process and collecting even more consumer data, Go’s underlying technology is a logical next step in Amazon’s brick-and-mortar retail pursuits.
Go is the latest foray in the company’s relentless push to minimize friction in retail transactions. “All the selection, all the convenience, low prices. And they get it to you in three days, then two days, then one day, then same-day, then same-hour,” says Mark Mahaney, a technology analyst with the financial firm Royal Bank of Canada. “Amazon’s done a lot of things to remove friction from the online shopping experience. The question is, can they do it with the offline store?”
The first Go shoppers on Monday were notably pleased, even entertained, with their experience. “It was a lot of fun,” Melody Coleman, a freelance user-interface designer who lives in Seattle, told me. “You could just grab stuff and walk out. It’s such a foreign concept to me ... It feels like I’m shoplifting.”
Once the novelty wears off, that shoplifting sensation will in theory become an appreciation of speedy transactions. In exchange for that convenience (which could also entice greater spending), Amazon will surely reap vast amounts of data. Amazon won’t say exactly what its cameras monitor, but it says the camera system is inspired by the same technology that guides self-driving cars. If the sophistication of Amazon’s retail sensors comes close to that of automotive systems that might produce four terabytes of data per day, then Amazon will surely gather insights more detailed than just shopping duration and items removed. Should a customer, for instance, grab a six-pack of microbrew beer, only to return it a minute later and exchange it for cheaper Rainier, Amazon might learn that the customer is a price-sensitive shopper and perhaps send them coupons for higher-margin items.
It’s easy to imagine Amazon Go’s technology in Whole Foods, which Amazon bought for $13.7 billion last year (a small section of the Go store already sells some products from the grocer’s Whole 365 line), but Sucharita Kodali, an e-commerce analyst at the market-research firm Forrester, says the camera setup could be too hard to scale, at least for now. A grocery store might be 100 times larger than the 1,800-square-foot Go store, and far more customers and products would need to be monitored simultaneously.
There are also issues with pricing. “Fresh produce, meats—a lot of that is sold on a weight basis, not a unit basis,” Kodali says. “I guess if you get to selling everything on a unit basis and standardize everything, maybe it could work. But you would have to significantly reshape your packaging and supply chain ... to make your pricing really consistent.” Likelier expansions, she says, would be bringing the technology to smaller Amazon retail outlets like its book stores, or licensing the technology to, say, fast-casual restaurants or gas stations.
Kodali is more confident that elements of Amazon’s technology will change the retail labor market—particularly for cashiers. “You won’t need as many of them, and the ones left behind will be forced to wear more hats,” she says. That’s not a groundbreaking notion; workers at Trader Joe’s have several duties that extend beyond the register. Yet Amazon certainly has incentives to further streamline its labor model. Labor is the biggest cost supermarkets face besides the food they sell, and grocers are already taking steps to make cashiers obsolete via self-checkout stands. Furthermore, brick-and-mortar success might be essential to sustaining Amazon’s swift growth, as 90.9 percent of U.S. retail sales still takes place in stores.
This has some worried about the economic implications of Amazon Go’s technology becoming mainstream. Cashiers are economically vulnerable—their median pay was about $20,000 in 2016—but numerous; 3.54 million Americans hold the job, second only to the number of retail salespeople. The repetitive nature of the job leaves it particularly susceptible to automation; researchers at Oxford University have concluded there is a 97 percent chance cashier jobs will be automated, and that was back in 2013.
Those who shopped at the Go store on opening day were pondering what the technology would mean for workers who hold low-skill jobs. Joy Carter, a service-industry worker and one of the masked protesters, said Amazon’s business model was “built on exploitation of low-paid workers,” adding, “It’s not going to be anything good for the working class.” (Amazon declined to answer questions about its new store, and did not respond to a request for comment about its pay practices.)
Others were more sanguine. Every non-protester I interviewed—those waiting in the block-long line were, of course, a self-selecting population—saw the replacement of cashier jobs as an inevitability of technological progression. Coleman, the designer, said low-skill jobs will be eliminated no matter what Amazon does; Andrea Guichard, a college student who drove about an hour to attend the store’s opening, said automation and machine learning will create more jobs than they destroy. Jeremy Ickes is a flagger for GLY Construction, the contractor that has built many of Amazon’s office towers. While guiding concrete trucks past the queued shoppers, he explained that his job wouldn’t be possible without Amazon. “They provide thousands of jobs, in multiple careers. It’s not just computers,” he said. “All these buildings have to be maintained. There’s construction all the time.”
Alison, a consultant who declined to give her last name, felt the protesters, like the city they live in, would also be swept up by Amazon’s growth. “I don’t know what they’re protesting,” she said. “They’ll likely end up working at Amazon in the future, too.”