After a year of teething trouble at the Amazon Go cashierless concept store in Seattle, the e-tailer has finally opened its doors to the public. While a fascinating exercise in the potential for technology to streamline choke points in a store, no amount of Jetsons-esque technology can prepare the company for a stampede of consumers accustomed to Flintstones-era retail.

Ready for prime time?
The Amazon Go store represents the pinnacle of current generation technology, described by the retailer as: “Made possible by the same types of technologies used in self-driving cars: computer vision, sensor fusion, and deep learning,” which allows consumers to “simply use the Amazon Go app to enter the store, take the products you want, and go! No lines, no checkout.”

However, the Amazon Go store has not had an auspicious start, with the pilot raising issues such as not being able to track more than 20 people in-store and incorrectly tracking items that are moved from shelves. While the bugs have apparently been ironed out, the idiosyncrasies of human behavior are likely to be just as problematic in unattended retail as they are in driverless cars.

The Amazon Go storefront
Bloomberg News

“The true test to see if this thing is ready for prime time is to let moms with 3-year-olds shop in the store,” said Michael Moeser, director of payments at Javelin Strategy & Research “There’s a big difference between a 30-year-old picking up and putting back items and a 3-year-old knocking over displays and throwing things in a cart.”

Tracking data, not sales
The reason for the complex array of sensors in the Amazon Go store relates to a number of factors — customer analytics, inventory management and real-time diagnostics — but there is also one critical reason for the technology: trust.

In a study conducted by the University of Leicester in the U.K., losses incurred through self-service technology payment systems totalled 3.97% of stock, compared to just 1.47% otherwise. There are checks and balances in place with today’s self-service checkouts, but these are far from foolproof.

A key difference with Amazon Go is every shopper is known from the moment they enter the store. Amazon has glass-gate turnstiles set up at the entrance, prompting shoppers to check in by scanning an app, much like subway riders might scan an app or a fare card before entering the system.

In Amazon's utopian shopping model, nobody is a stranger. This doesn't wipe out the possibility of shoplifting — any more than security cameras or guards wipe out shoplifting at other stores — but it does change the dynamic in a meaningful way.

“Amazon may be planning on using this one store as a final testing ground, as everyday traffic from the public will test the store concept’s technology and algorithms, and will likely lead to improvements,” said Dan Keyes, research associate, e-commerce at Business Insider Intelligence.

Amazon Go glassgate turnstiles
An Amazon employee scans in to shop at the Amazon Go store in Seattle. Bloomberg News

Robo-retail at scale
Opening the store to the public is a big step, but also a very restrained one. There may be some significant limitations to scaling the Amazon Go concept to anything much larger than a convenience store or quick-service restaurant, with the overhead of technology costs becoming a gating factor at scale.

“No doubt it works in the context of a convenience store,” said Thad Peterson, senior analyst at Aite Group. “The real test is scalability, and if scalable, at what cost? How many clerks/POS terminals can be eliminated to offset what has to be a significant install/management cost?”

However, one of the biggest hindrances for the cashierless store may be that it is just a little too inhuman.

The Amazon Go store isn’t entirely a walk-in vending machine — there are associates present for shelf stacking and food preparation, but they are very much relegated to the background. According to research by Retail Touchpoints, 90% of consumers say they are somewhat or extremely likely to make a purchase when they receive assistance from a knowledgeable store associate. Again, the type of store may dictate the degree of human interaction required — you are more likely to ask for consultation on a pair of jeans than a carton of milk, but the power of human interaction should not be underestimated by Amazon, particularly in a store so reliant on digital solutions to an analog experience.

What about Whole Foods?
Finally, it is worth asking — what is Amazon doing here? Its intention for retail domination has been clear and present for some time; even prior to the Whole Foods acquisition, Amazon operated a dozen or so retail bookstores throughout the U.S. These doubled as showcases for the company's technology, but otherwise don't look much different from the traditional brick-and-mortar booksellers Amazon once competed against.

Amazon might need to consider its strengths and limitations at some point. It's not unheard of for Amazon initiatives to flop. A few short years ago, Amazon debuted an aggressively priced mobile card reader called Amazon Local Register, and this product quickly frustrated early users and eventually disappeared. The Amazon Fire phone was an even more spectacular failure, even though much of its technology lives on in other Amazon products.

If the company's retail ambitions fail, it would affect the 89,000 Whole Foods employees who are closely watching their new parent company's next moves.

There are already concerns that Amazon’s reliance on technology is trumping tried-and-tested retail practices. A recently introduced Whole Foods platform called Order-to-Shelf has caused food shortages in stores, according to Business Insider, which said staff described the system as “militaristic” and “morale crushing.” Customers and staff are blaming Amazon for the problem, even though the system was implemented before the acquisition.

Amazon may need to tread lightly to alleviate concerns that its passion for Jetsons-era technology is not blinding it to the needs of the average Flintstones family.

Nick Holland

Nick Holland

Nick Holland is a senior analyst at PaymentsSource. He has previously held analyst roles at Javelin Strategy & Research, Yankee Group and Aite Group.