What does a cognitive retailer look like? Part 1

This article provides a basic overview on artificial intelligence and some initial ideas how a retailer can use AI to get competitive advantage.

What is artificial intelligence and why is everyone talking about it now?

Firstly, let’s explore what is new about Artificial Intelligence (AI) and cognitive systems, such as IBM Watson and why retailers are so excited about starting to use them. As you may know, the term AI has been in use since the 1940’s but it is only with the advent of cloud computing and the significant improvement in easy to use AI capability and reduction in price point that adoption is significantly increasing. Put simply, cognitive systems can do some of the things human beings can (such as reading and understanding speech), they can also do things human beings can’t, like reading millions of documents at a time (and remember them), stay up all night (every night) and not let emotion get in the way of judgement. AI systems can understand, reason and learn. They also allow you to interact with them in a more natural way.

So what can cognitive systems do that traditional systems cannot?

Let’s take a simple example. Imagine a non-cognitive system scanning an image of a dress, like this one (worn by my daughter Isabel):

It will register that image as pixels and perhaps can derive from that some attributes, such predominant colour. But it requires a human being to look at it and tag it with more information.

Now imagine a cognitive system scanning the same image. It will understand that image. It will register that image as a dress (not pixels), and by looking at 100’s of images of other dresses (and by being trained by a real person who understands fashion) it understands (without being told) that this is a formal, sleeveless, pleated, party dress. It can also find images of dresses that look similar and also what else is being worn by people who wear this type of dress. So, it knows most people wear high heels with that dress. Using facial recognition, a cognitive system can also tell the ages and sex of the people who typically wear it.  

Taking this a step further let’s assume the system has also access to unstructured information on social media blogs about the dress, for example it can read reviews about that dress. A cognitive retail systems take this a step further and is able to reason.  From those reviews, it can understand that this dress is a party dress, great for weddings, but not for the bride. By analysing the sentiment in the language used in those reviews it understands that dress is associated with glamour, sophistication and formal occasions. So it can start to apply reasoning to help you choose one option over another when given a question. In this case “I am going to a wedding, what dress would you recommend?”. 

Over time, when you teach a cognitive system rules it will learn to apply them through machine learning. It can then start to answer questions such as “What would go best with that dress?”. Just like a human if you make more information available, for example other unstructured information such as location and weather, cognitive systems can have opinions: “What should I wear to the garden party in Greenwich on Saturday?”.

If you now add to that the ability to interact in a more human like way, for example being able to understand the natural language of sentences perhaps through text chat or your voice then you are able to ask questions and get responses from a cognitive system in a natural way.  

What does a cognitive retailer look like?

So if these are the building blocks of cognitive systems let’s explore how they can be exploited to drive a retail business in new ways. Staying with our dress example, imagine your are the buyer for a fast fashion retailer. Some of your key decisions are about designing a new collection, choosing the style, colour and quantity to order.  You will base those decisions on inputs from your traditional sources, what happened last year in terms of sales by store by country, what fashion shows and trade magazines are telling you, you might follow a few blogs and subscribe to some trend predicting services. 

Imagine now being a buyer using cognitive systems that understand, reason and learn. They are monitoring all these sources and more for you in real time. They have learned from you the "language of fashion" and how to find predicting indicators for trends. Better still they can look at millions of pieces of unstructured data you would never have time to do. For example correlating weather patterns to past sales, scanning social media sites and analysing the pictures that people post, analysing the films that they watch and the celebrities they follow. 

A cognitive solution reasons like you do and so is able to make recommendations with confidence levels and a basis for it. Imagine an information pack that details automatically why it believes demand will be at a certain level, at a certain time of year in a certain country or city or store. This allows you time to focus on the more creative aspects of being a fashion buyer and perhaps allows you to book your orders with the manufacturers a few weeks earlier than you can today and with more confidence.

Now think how this retailer looks to a customer. Imagine using cognitive solutions to learn from your very best sales assistant and scale that to everyone who connects with your customers from stores, through digital channels and contact centres. That is going to result in brilliant customer service, personalised advice and a very happy customer.

 Conclusion

Cognitive retailers will unlock insights from data they already have and data they didn’t even know existed. They will plan, react and trade in ways that puts them in another league to their non-cognitive competitors. 

Look out for more articles to help you on your cognitive journey… it’s going to be fun!

Tony Maile

@Mailetony