Fewer staff, more data
Image by Alexas_Fotos on Pixabay

More than a year ago, Amazon amazed the world with its Amazon Go concept of a convenience store without staff or checkout. Much less visible to us are the many initiatives in China, even though several chains have already opened unmanned stores to the general public. Given their potential for saving costs and collecting tons of data, unmanned stores are bound to take a great flight. However, there are still challenges regarding the underlying technology and the level of convenience these stores offer to customers.


  • Amazon Go was first announced in December 2016, but earlier this year, it appeared that Amazon was still struggling with the technology (e.g. keeping track of more than 20 customers simultaneously) and could not rollout the concept. Now, the company is said to hire installers and markeeters to actually build and promote its cashierless shops.
  • com, the second biggest online retailer in China after Alibaba, has announced that it will open hundreds of unmanned shops in the coming years. JD uses technology similar to Amazon (a combination of cameras and facial and object recognition) and it will offer its technology to other retailers as well. Using similar technology, DeepBlue Technology has developed its Quixmart system and claims it will open no less than 100,000 stores in China within three years.
  • A simpler system, based on RFID tags on each product, is deployed by the likes of BingoBox and WeChat (i.e. Tencent). RFID may be easier to implement, but could be more expensive in the long run (a camera-based system only requires a one-off investment) and deliver far less data than camera-based systems (e.g. on how customers navigate to a store, which items draw their attention).
  • Alibaba and its subsidiary Ant Financial will offer technology and payment solutions to retailers who want to build their unmanned stores or others who may want to open unmanned (fast-food) restaurants, karaoke bars or even office spaces.
  • Taking matters one step further, MobyMart is proposed as a self-driving and unmanned convenience store concept developed by Wheelys (a Swedish startup selling bicycle-powered coffee carts).
  • In its most recent report on retail, PwC concludes that a majority of customers in retail location (68%) want to be able to check offers from other stores and 59% would like to see, and possibly order an extended range of products beyond a single store’s offering. This implies that consumers want to see some of the online shopping experience (of browsing and comparing products from all shops at once) reflected in physical stores and store staff might actually be a burden, rather than helpful, in this respect



Unmanned stores aren’t entirely new. Farmers have long used so-called honesty boxes to sell their produce without having to operate a proper store and (coin-operated) vending machines have been around since the late 19th century. In fact, vending machines of all sorts still continue to gain popularity for a growing range of product categories and contexts, including small-scale farmers, and they are increasingly coming to resemble small convenience stores (e.g. Bodega in the U.S.). Japan especially has a well-developed vending-machine culture.

The new generation of unmanned stores, as proposed by Amazon and several Chinese brands, is different in that they seek to mimic a more traditional shopping experience of browsing and touching a great variety of products

The new generation of unmanned stores, as proposed by Amazon and several Chinese brands, is different in that they seek to mimic a more traditional shopping experience of browsing and touching a great variety of products. Moreover, they try to make the shopping experience as frictionless as possible by eliminating the need to scan individual products (as with current self-service checkouts) and automating the payment. This reduction of friction also means that these unmanned stores represent a break from the broader self-service economy in which digital technology is used to shift tasks from a service provider to consumers (e.g. having to book a trip oneself instead of enlisting the help of a travel agency); these stores (in theory) can actually help consumers save time, while also saving on costs.

From a business perspective, there’s great interest in the precise behavior of customers in a store; how they move around a store, how much time they spend and which products they pick up and (more importantly) which ones they place back on the shelf. This is also why they (e.g. Amazon and JD) opt for a camera-based system that tracks individual customers throughout the store and keeps track of the products they place in their basket. The technological challenges of such a system are obvious; how to recognize individuals from a great variety of angles and how to track their purchases when there are constantly intermingling. The RFID-based systems are much easier to implement (these simply scan a number of tags upon leaving a store), but they will not also deliver the kind of rich data that cameras will. At this point, both systems are still far from perfect. Amazon is struggling with tracking large groups of shoppers at once and even the RFID-based solutions aren’t error-free yet. Also, equipping a store with tens, if not hundreds, of cameras and other sensors is very costly still and so is the necessary computing power (to be sure, adding RFID tags to each product could even be more expensive in the long run). Once these problems are solved and costs are lower, unmanned stores are bound to pop up across the globe.

The question remains whether all kinds of shops will be stripped of staff. This development seems obvious for convenience stores and supermarkets, where the majority of customers want to save as much time as possible. Less obvious is the automation of fashion stores and other shops where shopping is (to some at least) much more of a valuable experience in itself. Even though these kinds of shops could also transform and increasingly come to resemble the online shopping experience of quick browsing and comparing goods without any human interference, traditional retailers will hesitate to go down this road. It is much more likely, and Amazon and JD’s efforts underscore this point, that e-commerce brands will use unmanned stores as a physical front-end to their multi-channel product offerings. From their perspective, the unmanned store is a real-life webshop that adds a physical (and thus palpable and smellable) interface to their inventory.


  • Traditional brick-and-mortar retailers will struggle to adopt the unmanned store, not only because they lack the technological expertise, but also because they will have to radically reconsider their core-beliefs regarding the shopping experience and consumer preferences.
  • Unmanned stores will boost demand for high-end cameras, software and processors for facial and object recognition, other (biometric) sensors that can help keep track of customers and products.
  • As for the digital native unmanned stores such as Amazon, they will most likely be the sole owners of the data their shops produce. Those who are forced to procure turn-key solutions from others could be forced to share data with their suppliers. Those suppliers could be other retailers, dedicated tech companies or financial service providers (e.g. Ant Financial) who could build extensive customer profiles by combining shopping data with data about their financial status (i.e. knowing how much someone can spend and what he or she is interested in)