Is “Just Walk Out” retail a vision of the future?
Just walk out (of the shop)
Want to see the future in action?
Well, it’s happening now in Wembley, White City, Chalk Farm and Canary Wharf.
For these are the (slightly arbitrary-sounding) locations in London of the first “Just walk out” retail stores outside the USA.
Amazon Fresh stores replace the “unexpected item in the bagging area” machines with computer vision technology to completely eliminate the need for any kind of checkout process at all.
I went to have a go.
The Canary Wharf branch is rather out-of-the-way, in a new development that doesn't yet have much footfall.
You need to have the Amazon app on your phone and to scan in at the gate on entry. Interestingly, to make this work, Amazon have ended up introducing something that looks remarkably like a tube ticket barrier:
Once you’re in, you just walk round the shop, taking stuff off the shelves and putting it straight into your bag:
I did my best to confuse it. I took a tub of hoummous, put it in my bag, wandered round the shop for ages, took various other things and then returned the hoummous. I also wore my cycle helmet for some of my visit but took it off part way round:
When you’ve got everything you want, you just much straight out of the shop. There’s no need to scan the gate on the way out - it opens automatically and you’re back in the street:
It’s weirdly disorientating not going through any kind of payment process at all; though magically convenient.
And the key question? Did it get the payment right? Spot-on! With no hoummous!
The technology is largely the same kit that has been developed for self-driving cars: video cameras combined with machine learning to automatically figure out what is happening.
But in many ways, this is a much better application than self-driving cars. The problem with self-driving cars has been the edge cases; the circumstances in which the car encounters something its machine learning experience can’t process.
A car can have driven thousands of miles along ordinary roads building up a vast library of examples of what happens when things are normal but then finds itself encountering something new. What is this thing? How will it behave?
This is one of an almost infinite number of edge cases that a self-driving car needs to be able to deal with, safely.
By contrast, computer vision in a retail store needs only work in a retail store specifically designed for the purpose. The system doesn’t only rely on cameras; the shelves are pressure-sensitive. The camera is used to match up which customer has taken which product.
The Amazon Fresh shops using this technology in London are all convenience stores, but Amazon has just opened a full-size supermarket in Washington running the system - and there you don’t even need to use the app; you can simply wave your palm at the sensor to identify yourself.
It seems inconceivable that this technology is not the future of retail: the benefits to supermarkets are so great. There are few innovations that offer the magical combination of reducing the friction between customer and purchase, providing a stream of rich data and reducing operating costs. (Obviously, the latter point is dependent on this system working out cheaper than all the “unexpected item in the bagging area” machines; but once it’s scaled that seems plausible especially if, over time, computer vision can take over entirely from the shelf sensors).
Just walk out (of the bus)
So why am I telling you all this?
Well, partly because it’s a reminder of where customer expectations will be headed. Within a number of years (and I’m not going to attempt to guess how many), customers will be used to simply taking items from shelves and walking out. Given that parts of the public transport industry (especially buses outside London) have only just moved beyond cash, I’m sorry to tell you that the goalposts are moving again.
But it’s also indicative of a significant opportunity. The reason Amazon are the first company to roll out this technology at scale is because Amazon are the world leaders in reducing purchase friction. Throughout its history, Amazon has been obsessed with eliminating things that get in the way of a customer handing over cash. From “One-click” web payment to Amazon Prime, their constant focus has been on eliminating the barriers to a successful transaction. It’s not the only thing that made Jeff Bezos the richest man in the world (an obsession with costs that has often driven Amazon into distinctly dodgy employment practices has also contributed), but it’s a big part.
Well, public transport can also do with friction being removed.
Contactless payment was a huge step forward (and one of the most important transport innovations of recent decades) but it has limitations. It requires the customer to touch in and touch out. Fumbling for a wallet at a ticket gate can be stressful. The association with payment by waving a bank card may bring home the cost of the journey more than - for example - turning an ignition key.
A solution that enabled customers to simply walk on and walk off would reduce friction considerably.
Osnabrück, in Germany, is part-way there. They are currently testing a system in which customers check in on their phone as they board the bus; but can just step off and the phone recognises when they leave and automatically deducts the correct fare. But it still involves fiddling around with an app.
But a combination of facial recognition and computer vision could mean that registered customers can simply walk around the network without any need for any contact at all. You would need some form of ticket gates for either entry or exit (as per the Amazon store) but you wouldn’t need both. At a stroke, half the ticket gates in Britain could be retired - increasing capacity and dramatically reducing customer friction. And while I didn’t get to use the latest Amazon store with its palm recognition technology, not having to fumble for a phone or wallet would eliminate friction even further.
This isn’t the only option, of course. Switzerland uses mobile phone data to achieve a remarkably frictionless national travel app. Fairtiq is a standalone app that is also embedded in the SBB master app. With Fairtiq, you simply press the “Start” button when you begin a journey and then travel anywhere in Switerzland by any mode. Train, tram, ferry, bus - it doesn’t matter. At the end of the journey, you’re meant to press “Stop” but if you don’t, the app will do it for you. You then get charged the correct fare.
I’m certainly not an advocate for any particular technology; but I am an advocate for using technology to eliminate transaction friction.
The various Government strategies envisage contactless payment and a single ticket booking website.
They’re obviously progress on where we’ve been before.
But let’s learn from Osnabrück, Switzerland and - of course - Amazon; and build the future as opposed to catching up with the present.
