AI Vending Machine -- Future Vending Machine

WHY DOES COCA COLA NEED AI?

The Coca Cola company produces about 3% of all the beverages consumed globally. Its distinctive red and white logo is recognised by over 90% of people in the world.

But maintaining this leadership position takes hard work, with huge investment in marketing. A few years back Coca Cola was spending more on advertising than Apple and Microsoft combined.

It relies on huge amounts of data to understand why people are buying. It takes even more to figure out why they switch, and what makes them buy more. Nowadays, Coca Cola has an extensive product range. So gone are the days when success was mainly about getting people to choose Coke over Pepsi.

There are similarities with other brand-conscious purchases such as trainers and luxury fashion. There may be some functional differences between competing products, but sales seem disproportionately dependent on attitudes to brands.

R&D is key to Coca Cola’s success, not just in products, but also understanding customers and brand effectiveness. The volumes of data involved in the last two are immense. The relationship with buying patterns is complex, and is one of many compelling reasons Coca Cola uses AI.

AI is now fundamental to Coca Cola’s business, something it acknowledges publicly:

AI is the foundation for everything we do.

– Greg Chambers, Global Director of Digital Innovation, Coca Cola

Of course, there are also endless opportunities to use AI across the value chain, for example in distribution, logistics and manufacture. But a company like Coca Cola uses AI first and foremost to improve the customer and brand experience. Business functions that directly affect these are the biggest beneficiaries of AI.

WHAT IS AN INTELLIGENT VENDING MACHINE?

There are three different aspects to how Coca Cola uses AI in a vending machine. Each reflects one of the three main functions a vending machine fulfills:

• A physical sales outlet at a location with high demand;
• A contact point with loyal customers; and
• A showcase of different product choices.

Success of each function is measured by different things. Understanding the data to improve that is not straightforward.

For example, how do you choose the best location for a vending machine, and how do you decide when or if to move it? How do you persuade loyal customers to buy more drinks, and reward them for doing so? And how do you decide which drinks to stock?

In many ways, this is a microcosm of what Starbucks deals with for every store it opens and runs. But of course, at a vending machine there are no staff around to do anything needed.

So one way to look at it, is that an intelligent vending machine should much of what you’d expect from a store, but automatically. The moment you break that down into the data needed, AI starts to make some sort of sense.

COCA COLA USES AI: TO DETERMINE THE BEST LOCATION FOR VENDING MACHINES

Using the Starbucks analogy again, deciding on the best location for a vending machine is something that can be modelled. This requires using location data, such as footfall, local economic indicators, and of course competitor information. There’s also going to be a body of comparison data from other machines.

Once a machine has been set up, understanding sales data is key. It’s used to validate how the machine is performing as a sales outlet compared to expectations. For this, it’s not enough to simply know drinks sales in a given period. So connectivity is a desirable feature, allowing data about sales to pass directly to the company.

However, the power of what Starbucks and others do when evaluating a store’s performance is about much more than this. One key activity comes from more personalised insights into the transactions. This is where loyalty schemes and apps are key.

Which is where the second function of the intelligent vending machine comes into play. But it’s only possible because the humble vending machine is part of the Internet of Things. i.e. it’s now a connected machine.

COCA COLA USES AI: TO ENGAGE WITH LOYAL CUSTOMERS

An intelligent vending machine should integrate with a company’s loyalty programme. If the loyalty scheme enables payment, the machine should accept more than just cash payments. And this is where Coke vending machines start to get more interesting.

In Japan, customers who use the “Coke On” smartphone app collect points when buying from the machine using their phone. Coke gets all the usual loyalty scheme benefits from this, primarily customer preference and buying data. And the customers are rewarded with a cash value for their points against future vending machine purchases.

In other markets such as Australia, Coke trialled a different intelligent vending machine. I was again designed to encourage and reward loyalty while gathering customer and sales data. In this case, customers could use their phone to order drinks in advance, and collect them from the machine.

This kind of innovation isn’t just about connectivity — this time with a phone as well as the internet. It’s also about having digital displays on the machine.

With both connectivity and a display, the opportunity for AI steps up further. This time, Coca Cola uses AI to turn the vending machine into a fully-fledged “smart” device.

COCA COLA USES AI: TO SHOWCASE THE COKE BRAND & PRODUCTS

Vending machines have always been used as prominent signage points, promoting brands through posters. In recent times, these have become more sophisticated. Digital displays allow the use of a variety of high quality images and videos in one place.

The Coca Cola vending machines take this much further, because of intelligence in the actual vending machine. This concept goes back nearly a decade, to the company’s original “Freestyle” vending machines.

The first Freestyle machines allowed customers to decide exactly what drink they wanted from a wide range. The machine would create the exact chosen selection then and there. So a machine could offer up to 200 variations on the base products, mixing and dispensing each as required.

The new generation of Freestyle machines have a large touchscreen — bigger than most home computers — and of course a mobile app. But they also have lots of dormant features to be activated later, such as audio capability and optical sensors.

With this kind of hardware, connectivity and intelligence, the future opportunities for how Coca Cola uses AI seem only limited by the imagination.

The humble vending machine has become a smart device with the potential to rival anything in the home. And in the process, it provides a rich source of data for Coca Cola.