Amazon Go – A Winning Bet in the Convenience Store and Grocery Business

Amazon Go – a new architecture cooks up a storm in an industry stuck in time.

Early in December, 20166, Amazon announced its latest foray into the grocery business – Amazon Go – which relies on what Amazon calls “the world’s most advanced shopping technology” to provide a Just Walk Out shopping experience1. No lines, cashiers, and no credit cards are required at this store. But what makes this technology so advanced, and why have other retailers failed to bring an automated experience to consumers? What has allowed Amazon to Just Walk In to the brick-and-mortar convenience store business and cook up a storm? The answer lies in the company’s amazing ability to bring together existing technologies, from outside and within its own portfolio, and implement disruptive architectures in seemingly boring industries.

There are no radical or modular innovation elements involved in this “most advanced technology.” The stores use computer vision, used extensively in military technology, our personal computers, and now self-driving cars, to track customers as they move around in the store. Simple RFID and QR codes allow the store to know which customer has walked into the store and which items are being picked up from the shelves. Amazon applies its existing deep learning capability to improve the accuracy of its inventory tracking2. Let’s say if the RIFD sensors cannot tell whether you had picked up turkey or salami, then the store will draw on tiny amounts of computing capacity on the AWS infrastructure to analyze your purchase history and determine what you may have picked up from the shelves. The plethora of sensors across the store can then tell that you are walking towards the exit. As you pass through a “transition zone” computers add up your purchases and charge you through a pre-existing mobile payment platform2. No new technologies, but a beautifully orchestrated symphony between sensors and deep learning algorithms.

Video introducing the Amazon Go concept.

What is even more interesting is the failure of incumbents to develop a timely response. Amazon had actually filed a patent for this concept 3 years ago3. One could even argue that Amazon simply transplanted a tweaked version of its Kiva warehouse capabilities to the Amazon Go concept store. Amazon’s highly automated warehouses use the same sensor technology as Amazon Go to improve inventory tracking and ensure that operators pick the right items from shelves. Despite the concept being displayed in plain sight, the incumbents in the convenience and grocery markets simply could not respond. A case could be made that they should have seen how Amazon’s warehouse innovations could be brought into their own operating model. After all, most convenience and grocery businesses maintain at least some of their own warehouse capacity. This situation delineates the challenges that traditional businesses face in incorporating digital technologies. Convenience and grocery businesses aren’t set up to have IoT innovation divisions. Even if they attempt to partner with “technology-as-a-service” providers, integrating different pieces of the puzzle across platforms that may not be compatible without additional work or resources ends up proving very challenging. This situation also exemplifies how incentive structures at operating companies cause them to miss these waves of change, because they define a very narrow competitive scope.

Amazon is just beginning to unleash architecture-driven disruptions. The company has developed a powerful portfolio of technical capabilities, and its nimble organization structure allows product marketing, operations, and technical teams to work collaboratively on new problems. While they have not full embraced LEAN testing methods – often focusing on growth over demonstrating profitability first – they are very good at leveraging their resources to test new ideas that are natural extensions of their existing capabilities. In addition to the architecture innovations, Amazon Go will benefit tremendously from Amazon’s supplier base and infrastructure already developed for its Amazon Fresh business. This portfolio of resources creates tremendous threat for many industries, as Amazon can implement new architectures without having to worry about economies of scale in individual businesses.




Mark 43: Redefining Law Enforcement in a Mobile World


Flipboard – Started from the Top and Now We’re Here (in a Hole)

Student comments on Amazon Go – A Winning Bet in the Convenience Store and Grocery Business

  1. Way to spot the architectural innovation! I could envision other retailers — and even grocery stores — deploying similar systems; I don’t know the grocery business well, but something tells me people have a strong connection with places like Whole Foods or Safeway, and that these companies have a strong connection with food suppliers that will give them a little time to catch up. Even if Amazon cannot upend Whole Foods, say, maybe there’s an opportunity for the company to sell/license its “just walk out” systems to other grocers and/or retailers… I could envision Starbucks getting there soon on its own.

  2. Great analysis. I find it puzzling that incumbents have not been alarmed by this. I also wonder how Amazon tends to monetize. The margins in the grocery business are notoriously thin, and even though this type of technology would perhaps enable Amazon to run a brick and mortar grocery business with a slightly lower cost structure, it doesn’t seem likely to me they want to build out this type of business. Purely speculating, I would guess that they will roll out a few stores in order to collect data, dial-in their learning algorithms, and calibrate the technology. Later, they will probably seek to license out this technology to brick and mortar grocery stores (and probably other retailers as well). What do you think?

  3. It seems Amazon has relayed learnings from its’ warehouse business to grocery stores. This operational efficiency does not seem to exist outside of Amazon’s grocery store, comparing to stores such as Kroger, Safeway and Whole Foods.

    Decreasing time out of the shopping experience is equivalent to decreasing the pain of grocery shopping. Long, awkward and painful waits in the grocery checkout line have been eliminated.

    I agree with you that this outcome would not have been possible at an organization that does not enable new and disruptive approaches. Amazon seems to have gotten the operational side of things right.

  4. Thanks for a nice post on a very recent architectural innovation.

    Checkout lines are painful but they are not the worst thing in the world. As you mentioned, Amazon knows how to innovate and design products and services around customer needs with the latest technologies. In my opinion what we can expect for this initiative is that Amazon Go will continue to evolve in terms of products and services stores have to offer, in terms of algorithms they run on modeling the users (since they follow the user in store over their phone, and the user is already buying things online from them, Amazon will gain a competitive advantage over their pure web based competitors, ebay etc.). I expect Amazon to have a distinctive advantage to spot the location and size for the next store and what the sell there. Executed well, Amazon Go and its successors can provide substantial growth in groceries and meals to compete with old and new incumbents (Blue Apron, Hello Fresh) as well.

  5. Very interesting post, Kunal!
    While Amazon definitely leverages its Kiva warehouse capabilities with Amazon Go, I don’t think it’s enough of a competitive advantage to successfully run a supermarket chain. From a customer’s perspective, it is convenient not to queue and pay at a cashier, but location, product range and brands, price and quality seem more releavant factors for the choice of the supermarket.Thus, partnering with existing grocery chains seems like the more successful alternative to me

Leave a comment