Reverse-engineering Amazon Go: What you should know

February 14, 2018 

<< Previous

Recommend This Post

9

Sometimes you can learn more from thinking about why something works rather than just how it works –and that’s especially true for a new format like Amazon Go. We spoke with two industry experts who spent several days observing and studying the Seattle store, and reverse engineered their observations in order to understand why Amazon Go works so well for customers and how it’s defining a new competitive space in the retail ecosystem.

What is it?

It’s hard to categorize Amazon Go’s format. It’s been called a convenience store because of its size (1,800 SF of selling space, 2,400 total), but the product mix and shopping occasions served are very different from your typical convenience outlet located in a crowded, city center office worker environment – there’s no tobacco or hot coffee.

And, although the emphasis is on food, it’s definitely not a restaurant or grocery store. Instead, Amazon Go seems to be defining a new competitive space in the retail ecosystem.

Competitive advantage

Our reverse engineering produced this short but important list of three ways that Amazon Go has creates competitive advantage over other outlets that sell food.

1. When consumers use the store

Most stores are built mainly to serve either food-at-home or food-on-the-go but Amazon Go is built to serve both.

The assortment supports two eating occasions – grab-and-go and heat-and-eat. What makes this work especially well is that both options are available during the entire day, and it’s so easy to get in and out that shoppers give no thought to visiting a couple of times a day.

The store effectively serves as both refrigerator and pantry for regular customers, virtually eliminating the need to buy ahead. 

2. How consumers shop the store

Amazon Go is built for fast in-and-out visits: You can get in, buy a sandwich and a drink and leave in less than 30 seconds. The speed of the trip is enabled by two features:

• Most of the product is displayed on the walls in an easy to follow, logical order that aligns with the needs of most customers.
• Open space in the center of the store makes it easy to avoid being delayed by another customer, so there’s virtually no waiting. This is true even when a lot of shoppers are in the store.

Shopping is also faster because:

• Most orders involve just three or four items and the products are compact in size, so there’s no need for a shopping cart.
• All the products are displayed within easy reach, mostly within the knee to shoulder level “strike zone.”
• There’s no need to interact with staff expect for purchases of alcoholic beverages.

3. Low operating costs/high utilization business model

The business model behind this store directly addresses the main vulnerabilities of today’s self-service grocery store. It provides Amazon the flexibility and pricing that allows much lower prices, if necessary, without sacrificing profitability.

The efficient utilization of assets means they can attract capital with a competitive rate of return (even though Amazon has not historically needed to).

• Operating costs are kept low because there’s no need for most customers to ever interact with store staff, and backroom processing is driven by real-time signals about what’s selling.
• High inventory turns are achieved by setting shelf inventories to meet immediate demand and by continuously but unobtrusively restocking based on replenishment cycles as short as one hour.
• Asset utilization is kept high by encouraging a steady flow of business, because it’s so easy for shoppers to get in and out quickly. At this point, weekly sales per square foot are estimated to exceed $45, with plenty of capacity to go higher.

BMC POV

Our reverse engineering of this first Amazon Go store holds two closely related implications for those who are striving to get a glimpse of the future of food retailing.  In the future:

• High-performance retail units will be achieved by stores that are built to serve narrow, well-defined market segments. The main performance driver will be hyper-localized assortment, but tightly targeted value propositions will also be a contributor.
• Store performance measurements will depend on the target store’s customers and will be influenced by outlet design. In this case, the key performance measure is customer throughput per hour – this is not about increasing basket size, it’s about moving more paying customers in and out of the store.