Wharton's Barbara Kahn and Emory's Ryan Hamilton discuss the Amazon Go convenience store.

Even before it opened, Amazon Go began drawing deep skepticism over whether it could possibly work. Would Amazon be able to create a convenience store in which shoppers gathered their goods and walked out the door having been automatically charged for their purchases? The technology behind it — like driverless cars — has captured the public’s imagination. A reporter from The New York Times staged a shoplifting heist with a four-pack of vanilla soda around the time the store opened in Seattle last month, and failed. The Amazon Go technology worked, and he got charged for the soda. Others have, though, managed to outsmart the system.

But as shoppers gauge the gee-whiz aspects of the store and convenience of not having to wait in line, Amazon is sharply focused on the shopper. The company is always watching. More than 100 cameras in the 1,800-square-foot store are capturing shoppers’ every move to total up the purchase, but that’s just the start.

Amazon isn’t saying what its plans are for Amazon Go — whether the company will roll out dozens more soon, or exactly how the store serves a greater corporate strategy. But many say that by recording shopper behavior at such a detailed level and being able to analyze it, Amazon is hot on the trail of the Holy Grail of retail: really understanding why shoppers do what they do.

“The big enigma has always been why does a customer buy; what causes someone to pick something off the shelf, put it in their basket and pay for it?” says Marshall L. Fisher, Wharton professor of operations, information and decisions. “That’s the number-one issue in retailing. One obvious step is directing traffic to the store or website. But then once you’ve got the customer in your crosshairs, how do you get them to buy?”

The Amazon Go store samples in physical form the online realm from which Amazon has been able to extract increasingly detailed information about how customers behave. Says Wharton marketing professor Peter Fader: “To the extent that it revolutionizes retail, the idea here is knowing who is buying without relying on loyalty programs. But in addition to knowing who is looking at what, who is picking an item off the shelf and in what sequence — that idea of really seeing everything could have dramatic implications.” It could change the way stores are laid out, he notes, and it could change where a concierge person comes in. “I think that the data part of it could be the big breakthrough, but at this point it’s still icing on the cake.”

Amazon’s data-mining motives aside, no one should miss how Amazon Go evolves the shopping experience, particularly for the next generation of shoppers comfortable with technology and surveillance. Wharton management professor Daniel Raff happened to step into Amazon Go for a bite to eat not long ago, and called it a “high-end” convenience store, with not just items you’d find at 7-Eleven or the Circle K, but also prepared foods of the kind offered by Whole Foods.

“It was a very pleasant shopping experience, but actually eating the food was even better, and I think this shows the kind of thoughtfulness about the customer experience and tremendous trouble to which the company goes to getting that right,” said Raff. “People who have followed the development of the website will be familiar with this. It’s just a very customer-centric experience.” Raff noted that Amazon is applying to the grocery industry the same basic approach it used to revolutionize bookselling. It also gives a glimpse into changes Amazon could make to Whole Foods, which it purchased last summer.

“The big enigma has always been why does a customer buy; what causes someone to pick something off the shelf, put it in their basket and pay for it?”–Marshall Fisher

The Amazon Go concept is “consistent with what Amazon has been doing all along and what their real mantra is, which is to focus on convenience and make things as easy as possible,” Wharton marketing professor Barbara E. Kahn said during a recent interview on the Knowledge at Wharton show, which airs on SiriusXM channel 111. (Listen to the full podcast using the player above.) “What most surveys have shown is that the biggest pain point in physical shopping is waiting on a line to pay your bill after you’ve picked out all the items that you want, and so they have eliminated this annoyance from consumers’ lives.”

Closing the Deal

Knowing what customers do is easy. Why they do it often remains opaque. Customer surveys and other tools yield a certain amount of data, says Fisher, but with Amazon Go, “the new thing here is a more detailed look into the customer’s head, to see what influences them once they are in the store. I think this will be an iterative process. I don’t think they’ll have the final answer immediately, and it’s not like we know nothing now, but this could steadily increase understanding of customer behavior.”

Even small changes can have a big impact. Fisher points to a large shoe retailer that changed its practice from expecting customers to know their shoe size to offering to measure them for size. That one addition upped the close rate. The theory was that the act of measuring engendered a little more interaction between the sales associate and customer, he said, and that was good for sales.

“The close rate in apparel is 20% — that is, 20% of people who enter the store leave having bought something, and even increasing it to 21% or 22% is a huge deal for a retailer,” says Fisher.

While Amazon Go doesn’t have cashiers providing personal contact, it does have people in the store offering assistance — a small corps of concierges to guide customers and answer questions. It’s hard to know whether Amazon Go augurs badly for grocery store jobs, but a human presence is clearly important. “There will still be roles for people,” said Kahn, pointing to the number of employees on the floor at Amazon’s physical bookstores. “A lot of people go to the store for that social interaction. Even in the bookstore, the people they hire come up and they talk to you and help you and initiate conversations about books, because part of the benefit of this is that social interaction. I don’t think they are going to eliminate that.”

But there’s no doubt that the Amazon Go experience is unique from the start. In order to be admitted to the store, shoppers must have an Amazon account, a recent smartphone, and the Amazon Go app, which leaves out some shoppers. There are no shopping carts. To enter, you swipe your smartphone, which opens a turnstile.

The app tutorial explains the concept like this: “Our checkout-free shopping experience is made possible by the same types of technologies used in self-driving cars: computer vision, sensor fusion and deep learning. Our Just Walk Out technology automatically detects when products are taken from or returned to the shelves and keeps track of them in a virtual cart. When you’re done shopping, you can just leave the store. Shortly after, we’ll send you a receipt and charge your Amazon account.”

“I think that the data part of it could be the big breakthrough, but at this point it’s still icing on the cake.”–Peter Fader

Having a physical store promises the company a laboratory for observing behavior that’s hard to watch online, where Amazon in 2017 accounted for an estimated 18% of online grocery and beverage sales, the largest share in the industry. While Amazon has been good at collecting click data on sales and buyer movement, what the company doesn’t capture as well are the details of why an online customer might look at merchandise and not buy it, says Paco Underhill, founder and CEO of retail research company Envirosell and author of Why We Buy: The Science of Shopping. With Amazon Go, “one of the things you are going to start pulling out is the number of people who walk into the canned soup section and go out with nothing. And understanding where you are losing is really critical to understanding modern retailing,” he says.

In terms of seamless transactions, the U.S. trails the rest of the world, says Underhill. “You go to a Korean subway station and there is a convenience store with a digital menu board where you punch things in and the items are delivered to your home hours later,” he says. “You go to Western European markets where merchants are struggling with unions and there is an incentive for them to reduce labor costs on the floor, and supermarkets often have scan and checkout technology far beyond anything we have here.” In the U.S., he said, merchants don’t have the same incentive because of low labor costs, but projecting into the future the labor costs in high-digital-literacy markets are probably going to go up. “If you look at the difference between Seattle and Walla Walla, Washington, you can see where some of that is going. Amazon is looking at other parts of the world and saying we should be playing with this, too.”

What’s also valuable in Amazon’s increasingly substantial foray into the physical realm is learning lessons for taking retail to places beyond the shopping mall and Main Street — for instance, airports, sports stadiums, gyms or corporate campuses. Says Underhill: “How do I find people as opposed to people finding me? What are the frontiers of modern shopping? Part of what the Amazon model here is trying to work out is to see how they are going to get to whatever the next generation of retailing is, which is pretty exciting.”

Where Amazon is heading with Amazon Go might have some significant parallels with the way the Book of the Month Club figured out how to make a great deal of money in publishing, an industry that has often struggled with thin profit margins. Raff has studied what made that adaptation so successful, and while tapping into (and cultivating) middlebrow taste was part of it, what really generated profits was being able to figure out the supply-chain issues. The BOMC not only did its own printing of previously published books, but also avoided over-production by asking subscribers in advance which books they would be interested in, and then printing more or fewer copies in response.

It was pretty sophisticated stuff for the time, Raff notes. “The Book of the Month Club made money hand over fist in its first 20 years. In a thin-margin industry, it looked like Silicon Valley in the 1990s.”

“What most surveys have shown is that the biggest pain point in physical shopping is waiting on a line to pay your bill after you’ve picked out all the items that you want, and so they have eliminated this annoyance from consumers’ lives.”–Barbara Kahn

According to Raff, Amazon Go appears to be about studying consumer behavior carefully, as well as using that information about when things are selling and where inventories are going. “I bet they are devoting enormous quantities of their sophisticated analytics talent to studying how to run Whole Foods more efficiently,” he said. “It’s pretty much commonplace in groceries that you draw people into the store with fresh food and fruits and vegetables and hope that they buy higher-margin stuff before they leave. This is the kind of thing Amazon is good at — analyzing demand patterns and optimizing the supply chain and so forth.”

The Walls Have Eyes

If customers are getting an easier, more pleasurable shopping experience and Amazon is getting a gold mine of data on customer behavior, who is getting the better deal? Having your movements tracked by cameras, after all, means trading away not only privileged personal information, but also privacy.

To some, it’s a win-win proposition. “If the close rate is higher, that’s an advantage to the retailer, but also for the customer” says Fisher. “What bigger waste of time is there as a customer to have a need, travel to a store and then walk away empty-handed because you couldn’t find what you wanted? It’s a huge hassle and frustration. The only time the retailer benefits is if the customer benefits.”

But others see dark clouds gathering to the extent to which personal data is being gathered and aggregated. “I suspect that the level of unease a shopper has being so closely monitored is correlated with age,” says Maurice Schweitzer, Wharton professor of operations, information and decisions. “We have gotten used to making a trade-off between convenience and privacy. For example, should I allow Google maps to know my location? It makes sense that retailers are taking this a step further, because the ability to collect, analyze and use data about customers has gotten so much better.”

“The close rate in apparel is 20% — that is, 20% of people who enter the store leave having bought something, and even increasing it to 21% or 22% is a huge deal for a retailer.”–Marshall Fisher

For some, the new levels of monitoring are creepy — like a stalker. For others, these new levels of monitoring are just another incremental step in our loss of privacy and a worthwhile price to pay for greater convenience. “I suspect that most consumers are marching along without much regard for how, just as professional athletes are monitored when they are on the field, we are monitored as we go about our daily lives,” Schweitzer noted.

Adds Underhill: “I do think there are some privacy issues there, and our ability to collect information has gotten way past our perspective to legislate.” Part of the success in alleviating the concern any customer may have about infringement on privacy is in not letting the observed feel observed.

How did it feel to Raff on his recent visit? “Certainly, your attention is focused on what is available and what you want. The seamlessness makes all of that very direct and everything else fades away,” he said. “What I do remember is that it didn’t intrude on my consciousness. I noticed there were cameras, but don’t think I took in at the time that they were anything other than ordinary security cameras.”

And how was the food? “It was just good, tasty food,” he said. “And how often do you step out of the office for a quick lunch and come back thinking that?”