Last year, Amazon opened a physical bookstore in Seattle, a program it has since expanded to other US cities. And this week, Amazon has announced that it is beta testing a brick-and-mortar convenience store in Seattle, a concept it has dubbed Amazon Go. While at first glance it is surprising that Amazon would be so eager to enter the world of physical retail, upon closer examination Amazon’s moves start to make a lot of sense. In truth, despite being best known as an internet company, Amazon’s remarkable success has always relied upon the company’s unique status as a leader in two seemingly very disparate milieus – the ethereal world of e-commerce, and the physical world of logistics. As I have written before, however, success in the digital era increasingly depends on developing business models and competencies that take advantage of the merging of the physical and the digital. Few companies do this better than Amazon does.
Aside from the sheer irony inherent in fact that the company which has done more than any other to disrupt the world of retail by helping usher in the era of e-commerce is opening a physical store, the novelty of Amazon Go is that Amazon has completely done away with the check-out line. Amazon claims to have completely automated the checkout process, by using what it calls “Just Walk Out” technology (which the company says employs similar technologies to those used to operate self-driving cars), so customers can simply grab items off of the shelf and walk out of the store, and they will automatically be charged the correct amount through a smartphone app.
Cashier jobs are not about to disappear overnight
Some commentators are already tying this news in to the broader concern in society about the threat that technological advances will result in widespread unemployment. While there are valid concerns and important unanswered questions about what the impact of recent advances in the automation of both white and blue collar occupations will be and how societies should respond, I don’t think that is the most relevant aspect of this specific development. Amazon’s announcement will not have a measurable impact on the ongoing trends towards increasing automation. I believe that this announcement is far more illuminating when viewed from the perspective of what it says about Amazon’s approach to customer data and its data privacy implications.
Yes, the job of cashier can be automated – this is nothing new (have you ever used a vending machine, let alone a self-checkout counter?). But there is a big difference between what can be done and what will be done. It is an age-old mistake to assume that decisions (corporate, government, consumer, etc.) are determined primarily by what is technologically possible. For companies to aggressively pursue automation, first and foremost there has to be a strong economic logic. And second, the company has to be comfortable that automation will not have a negative impact on the customer experience or the company’s brand. Replacing cashiers with technology is very different from doing the same to factory workers on a production line – not only is the economic logic usually much stronger in the case of skilled manufacturing workers, whose wages are typically a lot higher than those of cashiers, but their jobs, though they may be highly skilled, tend to be more precisely defined and predictable, and therefore easier to automate. Finally, factory workers are invisible to the end customer, so the company risks no negative impact on the customer experience caused by the removal of a key personal interaction that many shoppers view positively.
Even assuming the trend towards increased automation of cashiers accelerates, the impact will still be relatively gradual. Automated checkout kiosks have been around for a while now, and yet the total number of cashier jobs is still growing, albeit at a slower rate than the rate of growth in the overall labor force. Let’s not forget that ATMs were first introduced in 1969, and yet the number of human bank tellers continued to grow for decades, only reaching a plateau within the past decade, even as hundreds of thousands of ATMs were installed at the same time. So, like bank tellers, cashiers are unlikely to disappear any time soon, but the nature of their jobs will almost certainly change significantly as ever more sophisticated technology and automation are introduced into retail stores.
It’s all about the data
The point of Amazon Go is not primarily to explore ways of reducing the cost of operating a physical store by removing labor. Even in a context of high minimum wages and the falling cost of the sophisticated equipment required to make this experiment work, one suspects any cost savings (if they exist at all) would hardly be worth the trouble. If eliminating the cost of operating a physical store were the primary focus, why bother with a physical store at all (after all, this is Amazon we’re talking about)? Even if this pilot succeeds and is expanded, sales from physical outlets are unlikely ever to rival the importance to Amazon of its e-commerce or cloud computing services. Amazon is clearly interested in more than just disrupting the in-person convenience store shopping experience here.
On the other hand, you couldn’t design a better way to observe how consumers behave whilst shopping in a physical store than Amazon’s pilot program. As good as Amazon is at collecting and analyzing consumer data from its web and mobile operations, the physical aspect of shopping behavior has always been an important missing piece of the puzzle. Not only will shoppers in Amazon’s store will be carrying around smartphones that are logged into an Amazon app, they and the products they peruse will be subjected to an unprecedented degree of surveillance from sophisticated cameras and sensors. Does a shopper hesitate with their hand over one item, only to ultimately pick a similar, lower-cost item? What does this tell us about their behavior – is it idiosyncratic, or if multiple shoppers behave similarly, is it an indication that changes in advertising, product displays, or pricing strategies might result in more shoppers picking the higher-margin option? Amazon will be able to gain further insight when it marries the in-store data with what it already knows about shoppers’ online shopping behavior and purchase history (which it will be able to do in real time), to say nothing of data collected from other Amazon services, such as its streaming video and music services.
Quite simply, Amazon’s forays into physical retail sales will give it access to a trove of customer data hitherto unprecedented in terms of size, scope, granularity, and accuracy. To give just one example, security firms are already able to derive meaningful insights about people’s behavior by using artificial intelligence to analyze video footage taken from high risk public spaces such as airports, and they do this working with source data that are inferior by orders of magnitude to the data that will be generated by the cameras and sensors in Amazon’s retail outlets. By enabling them to combine detailed real-time behavioral analysis with a customer’s purchase history and personal preferences, developments like Amazon Go could confer enormous power on retailers. Once you start thinking in these terms, it rapidly becomes evident that the potential value of these customer data to Amazon will far outweigh any efficiency gains from automating the checkout process. What this means for data privacy, however, is another question entirely – one which was left unaddressed by Amazon in its announcement.