AI & Retail: Teaching to Learn

Artificial Intelligence

It’s a concept that has been around for longer than most people probably know. As early as 1950, Alan Turing was asking “Can machines think?”.

Back then the answer was very easy – no they cannot. They could barely remember a number larger than 10 without punching it onto cardboard.

A lot has changed since then of course. Just as Moore predicted with his law, our processing power increases have resulted in a device on my wrist having 1,000 times more processing power than an Atari 2600.

Additionally, the goalpost of what we mean and understand to be true AI is in constant motion. Our definition of AI has drastically evolved since Turing’s time, and it is still quite unsettled.

Is it a machine that can beat a human at chess?
That happened a long time ago with IBM’s Deep Blue chess-playing computer (which was, by the way, awful at poker).

Is it a machine that can react faster than a human?
That also happened a very long time ago. The industrial revolution introduced cotton machines that increased efficiency by 1,000s of percentage points. (Yet none of the machines could catch a baseball…).

Is it a machine from the future, programmed to remove the greatest commander of the human resistance, before he was ever born?
Well, not yet at least (if you forgive the time paradox issue).

It may be a difficult – or perhaps impossible – task to define AI. Yet, what is interesting is that the more we try to define it, the more we seem to be surrounded by it, completely oblivious. For instance, AI is working in the background when:

• Traffic lights change their timings based on traffic density and time of day.
• Advertisements appear on our daily web browsing that is eerily targeted to our wants.
• Emails, messages, and documents want to write themselves for us.

All very subtle alterations to the norm, but not drastic leaps of technology. Despite its subtlety of it, there is no doubt that AI is affecting our lives, and mostly for the better. Our understanding of how it can help us, and of how it can increase our quality of life, is becoming closer to mainstream and further from science fiction.

Regardless, of how clever AI becomes, there is one fact that remains – it will need data.

Data is what drives AI, empowers its proposals, and guides its decisions. You and I are the targets of its inquisitiveness, and if it is to truly help us it must know us.

Alongside the development of AI, we have introduced a real problem, in that our rate of data creation has grown exponentially. It is true that there has never before in the history of the world been so much data. While you read this sentence, the US alone added 167 721 Gigabytes to the internet.

Without AI, this data would increasingly just become noisy garbage. Our usage of it would be so tiny as to be largely pointless, meaningless, and valueless.

If I had to define AI, I would say that it is our collaborative partner in making sense of the world around us, in a time when the information is overwhelmingly huge for the human species alone.

So what does this mean for our business domain, Retail?

The cycle of a successful retailer is a relatively simple one.

More customers, having a better experience, more frequently.

This cycle in itself can be a huge resource of data. But it can only be a valuable resource if the data is well managed, meaningfully comprehended, and intelligently acted upon.

A silly illustrative example. In the corner shop in the village I grew up in, the shopkeeper knew every single one of the locals, to the extent that he could order stock knowing the individual villagers’ needs and habits. He even once demanded I pay for the candy bar that my little brother stole three weeks prior (I didn’t pay, but my brother did thanks to my persuasion).

Fast forward 30 years to a retailer that has a larger and more diverse customer set.

Stock management and assortments are not at all personalized to individual behavior

The standard approach is to fill the stockroom with more of the products which are in the mean average societal wish. This is what we call an informed gamble, hoping the statistics are correct and that they are relevant for your customers. At best, it is a shotgun approach that will certainly capture a lot of people…but not everyone.

Prices are reactive to supply and demand

Pricing starts as a wild guess, usually related to the cost and some idea of what a customer will willingly pay. But again, although the cost is generally a knowable variable, the customer’s willingness to pay a price is hugely varied, not only person to person, but depending on the age, personality, or even current mood.

Making a customer feel welcomed and appreciated is generic or too late in the transaction

We all aim for increasing customer loyalty, usually by trying to give the customer a feeling of being recognized, welcomed, and appreciated in your business. Our village shopkeeper had a smile and a greeting that is unique to every customer (especially my little brother). But larger retailers can usually only make a broad attempt at this, one that often falls short, and rather makes the customer feel like embossed numbers on a customer card shown at the register or just an email address.

…It doesn’t have to be that way

With data and its trusty partner AI, these points, along with many more, become situations that can be altered. With the right knowledge of your customers and with the incredible analytics available today, every retailer can offer that village corner-store customer experience, to everyone. Customers can be welcomed in a meaningful way, and importantly, in a way that doesn’t feel artificial.

I use that last word intentionally. Although Artificial Intelligence is a fantastic tool, if used poorly it is still obviously artificial. And if there is one thing humans are incredibly skilled at, it is to spot the artificial. For example:

• An email sent using the customer’s first name is easily spotted as a template circular. A ‘dear valued customer’ can sometimes be a more accepted approach.
• A personalized campaign that offers nothing the customer likes, is not personal.
• A smiling ‘welcome to the store’ staff member at the door, can be a mild annoyance in some cultures and is not personal.

If you get it right, and get to know your customers; their needs, likes, and dislikes, and you meet them…well, then they will smile, come back, and feel that human connection that drove retail forward until relatively recently. People will fall back in love with Retail again, enjoying the shopping experience that makes it, unlike any other industry.

Retail is becoming increasingly about fine margins. Ensuring that you not only cater to everyone so that they become and remain customers, but also that you cater to their entire family and their changing needs. Statistics aren’t great for this. You need predictive sciences, with analytics and AI-powered learning.

So as the world gets bigger, and we all feel less and less connected to each other, ironically it might be the cold dystopian figure of AI that comes in to enrich our lives, bringing us closer relationships in retail, that would otherwise have been left in a long lost and longed for yesterday.

Author