Vending machines should be the Shopify of physical retail

12.18, Thursday 24 Jun 2021

Here’s a man in Japan who sells home-cooked curries from a vending machine (YouTube, 2 mins 35 secs). I like this quote in particular: I have been running my vending machine business for about 40 years.

(Thanks Andrew Eland for sending this video my way.)

For a few years I ran a tiny bookshop in a vending machine, so obviously I am super into them (here’s a list of more), but the reason why is related to what the curry cook in Japan said: a vending machine is a minimum viable retail business, neatly packaged and tied up with string. A shop in a bottle.

I would love to see neighbourhood vending machines by busstops and in the corners of coffee shops, maintained by local individuals with handy essentials (battery packs, masks, sanitary products) and their own creations and hobbies: home cooked meals, soft toys, second hand books… really anything you see a friend starting as a side business on Facebook.

But it’s not easy, and I learnt a lot about what could be better.

Inventory management and operations

A major cost of a vending machine is the time taken to check it and restock it.

By connecting my machine to the internet (and a custom back-end), I automated inventory tracking – but two manual processes for me were setting the prices in the machine, and creating the “shelf talkers”. That’s the message underneath each item which has the marketing copy. It would be neat to have these in e-ink and set remotely from software, instead of having to print them each time.

There are some additional possibilities here: creating the “planogram” (the physical layout of the merchandise on the shelf) should be automated. Items at eye level sell better, so knowing about turnover and margin would help optimise the machine.

So inventory management is basically software for known workflows and a robust Internet of Things platform.

Metrics and distribution

Look at the way e-commerce works, particularly the idea of the funnel. Top of funnel, users are attracted from all kinds of channels: Facebook, ads on the tube, and so on. They’re moved towards purchase with marketing messages, and this process is measured and the messaging tweaked. Finally the purchase flow is also instrumented: do more users complete the purchase when the “Buy Now” button is at the top of the page or the bottom? You bet that the e-commerce site owner has tested that and they know.

None of this exists for vending machines.

Yes there are metrics: it would be awesome to see a dashboard of the footfall around the machine, the percentage converted to dwell within 3 feet of the machine, the percentage that starts hitting buttons to select an item, and so on. This could all be done with privacy-preserving sensors, and it would help to optimise locations.

But mainly I mean distribution: if a shopping website could only entire potential customers if they had already come to the website, it wouldn’t last very long. That’s the state that vending machines are in today. So how can they cast a wider net?

My attempt was to have the machine tweet every time it sold a book. That was a way for people, hopefully nearby, to get a timely reminder about the machine so that they would make a visit.

Better would be some kind of app that would let you see what vending machines are nearby, and pre-purchase items. How great would it be to go “oh, I need a battery pack” – then purchase it from the app, and it tells you what machine to pick it up from on your commute home.

So an integration into transit apps or Google Maps would also be useful, but being able to “follow” your local machines on Twitter and other social media feels like the more fun way to do it.

And then is it possible for the relationship with high-value customers to be retained and perhaps they could be offered a coupon if they haven’t visited for a month or two? All doable with e-commerce; that kind of detailed calculus is common. With vending machines? It needs some imagination.

Technology and integration

Ultimately I wrapped up the machine because everyone moved from cash to cashless, and replacing the coin changer with a card reader was going to be a whole bunch of work that I wasn’t in the mood for. Vending machines are pretty modular inside so the technology is ancient and fiddly but not a huge pain – it’s mainly that a card reader means having a merchant account and a monthly fee going to an old-school payments company. No easy Square readers here.

Really what you want is Apple Pay, loyalty points, reserved items, discount codes, upsell offers, all of that good stuff! But you’ll have to build it all yourself.

So none of what I’ve described above is insurmountable, in itself, but cumulatively it’s out of reach of the kind of people who would run independent vending machines – just as running an indie online store was out of reach for creators until Shopify came along.


Here’s a stat for you: About 44% of new food businesses started since the first lockdown are home-based, says BBC News.

I have a friend with one of these new food businesses. She sells soup. There’s a lot of leafleting and delivery going on. Like the curry guy, how great would it be if she could also have a vending machine sitting somewhere nearby?

Look, I don’t believe there’s a venture scale business in operating vending machines directly. The margins are too low and the operations cost is heavy.

But what we’ve learnt from e-commerce is that it’s a game of a thousand micro-optimisations. You measure every flow and improve each step just 1%… and added up, that’s the difference between loss and profit.

The current ecosystem is in the dark ages comparitively. Machines are expensive to buy and expensive to customise with software and poorly integrated. That’s frustrating.

Also frustrating are Amazon Lockers. Imagine having that amount of internet-connected robot real estate in great locations, and not providing a way for merchants to promote and sell their own goods to the neighbourhood!

So I do think there’s a startup in re-inventing the vending machine and becoming the Shopify of physical retail. Not one giant store but, like Shopify, a network of a million tiny ones. You’d create the machine and the software, and bring everything we know from online retail to physical, unattended retail, providing that as a platform to independent creators and nascent shopkeepers everywhere.

Tell you what, I’ll be an advisor if you start it.

Auto-detected kinda similar posts:

If you enjoyed this post, please consider sharing it by email or on social media. Here’s the link. Thanks, —Matt.