It has been said every age sees every idea humans are capable of conceiving. The profundity of the aphorism is apparent in our industry. Self-checkout was supposed to usher in a utopia of frictionless retail. It has been tried before and had failed. A self-checkout register was built and patented as early as 1984. Kmart introduced self-checkout lanes in 2001, only to let the idea quietly die in 2003
Chances are, self-checkout will fail again. Customers resent being forced to do work retailers are paid to do. What’s more, leakage (measured as non-scanning of items) – both deliberate and inadvertent – hovers around the 0.44% mark, measured as percentage of all self-checkout sales. The number might seem small, but to put the figure in perspective, leakage from self checkout lanes account for 9.5% of all store shrinkage. Some estimates peg the loss figure to an astounding 4% of all sales. And of course, this is an industry where margins range 1-4%. There is something about gargantuan companies begrudging cashiers their minimum wage that ticks off even those who aren’t habitual criminals. It’s a truism if a system can be gamed, it will be gamed. In addition, there are real glitches with the technology, and oftentimes a shopper pressed for time, will utter an expletive and walk away with an unbilled item.
The prime gripe is that self-checkout solutions, in their commonest contemporary avatar, defeats the purported benefit of self-checkout solutions, their raison d’etre, which is to save time. Internet forums are full of frustrated shoppers letting off steam, recounting efforts at scanning a cart full of supermarket products that go on for much longer than necessary.
There have been a number of rollbacks. Of course, rollbacks do not always make the news. Success received far more press than failure. But we at Litmus7 have been privy to a few of the latter. Wildly optimistic predictions were made about the number of self-checkout lanes in the past – 200,000 by 2007, and 325,000 by 2019. None of the fantastic numbers materialized.
One might think an idea with a certain amount of momentum gets perhaps too big to fail, and will eventually see some traction. To those we say – anybody remembers the Amazon Dash button? They are officially no more, with sales discontinued in March 2019. Contemporary retail is an arms race of shaving seconds off transactions, taking ease of use to ridiculous levels. It stands to reason to some ideas with fail. It is therefore important to curb your enthusiasm and not see every new toy as the harbinger of revolution.
Leakage can be reduced, and the user experience much improved. Self-checkout technology has to go from the current norm of “scan and go” to an Amazon Go level of automation, which is “pick and walk away”. But the surveillance measures necessary – a complex of cameras, sensors, facial recognition, image processing – are positively Orwellian; and what’s more – expensive. Whether customers will be outraged by such surveillance measures is an open question. Thus far consumers have been cool (who thought we would usher in 1984, out of our own free will, for a shot at impressing friends with our breakfast choices). The more pertinent question is the cost of these technologies. The company in the best position to offer an economic option for “pick and walk away” is of course Amazon, but even if the Seattle giant were to license the technology, would retailers bite? Retailers have been moving away from AWS, and it seems unlikely they would want to handover checkout to the industry hegemon. Startups are trying, but nobody has hit the right spot yet.
This analysis, and whether self-checkout will pick up, pivots on whether Amazon can manage the conflict of interest inherent in licensing its technology and/or if a third-party (a Google or a dream team of machine learning wizards from the Bay Area or Tel Aviv) can build viable technology in time. The former is unlikely. The second could be stymied by Amazon patents. Amazon filed for a patent titled “detecting item interaction and movement” in June of 2013. The abstract sounds eerily similar to how Amazon Go works. By the way, the patent expires in 2034.
While on the subject of patents and Amazon, it would be useful to revisit another patent – the single click checkout – granted in 1999. Many credit Amazon’s rise to the seminal piece of work, which Barnes and Noble tried unsuccessfully to replicate. If a patent saw Amazon through its first two decades, another might see it through the next two. But whether Amazon will license, and if they do, will retailers grudgingly accept the potential labyrinthine maze of issues remains to be seen. The future of retail could pivot on this very dynamic.