Skip to main content

What’s next for AI in retail?

(Image credit: Image source: Shutterstock/Maxx-Studio)

It won’t be news to anyone that AI is “the next big thing.” We’re constantly promised that AI will improve almost every aspect of our lives: doing our taxes, keeping our fridges full, travelling, education in schools – the list goes on and on. However, these promises are not exclusive to technology companies. Many retailers are now pushing the AI agenda, experimenting with new technologies not only to benefit their own operations, but also to improve their customer experiences. The questions we should ask now are how they are actually doing it, and how they could be using AI more effectively. 

Taking prediction one step further

Whether consumers are shopping online or in-store, the shopping experience seems to be the current playground for the AI players to call the shots. When AI meets the shopping experience, the goal is to predict what you buy – and when you will buy it. Prediction can drive everything from personalisation to fulfilment optimisation and all things in between. But the reality is, while AI can cut through the noise of an overwhelming amount of data and uncover the most important factors that should be applied when calculating specific consumer behaviours, retailers far too often assume there’s already a ton of usable data out there.

Good AI-driven prediction separates data that is noise from data that actually contributes to a better result, especially when it comes to personalised product recommendations. Basic algorithms can easily predict that black gloves will be nicely paired with a black jacket. But do you really need artificial intelligence to suggest to your consumer that two colours match? Truly thought-provoking, data-driven AI will predict what jacket is best for the consumer based on variables such as location, time of year and predicted fit. And then it will throw in a pair of suggested gloves before the customer reaches checkout. With personalisation, predicting which items to show a shopper next can generate a lot of value, even if it’s right only 50 per cent of the time.

AI remains in the experiment stage

Judging by the number of times the words “artificial intelligence” get tossed around in sales pitches, presentations, technology descriptions and company product pages, you would think they describe a mature capability well on its way to being rolled out across every retail enterprise. The reality is different. Gartner reports that a mere 2 per cent of retailers have already invested in and deployed AI, and 24 per cent of retailers are “experimenting” with it.

So what’s next? We know some of the love for AI comes from shiny object syndrome – if AI is the next big thing, then it must be the thing that solves all retail problems. However, retailers have to start by separating different kinds of AI from each other. There’s activity within natural language processing – chatbots and AI’s that can write product description copy, for example. There’s activity going on with image processing – recognising the difference between trousers and shorts, or assigning attributes to images that can be used in recommendations and other predictive customer-facing interactions. Either way, some of the most promising applications of prediction are in merchandise planning.

For complex retailing scenarios, AI can equip retailers with the insights to stay ahead of the curve every season. For example, AI can help retailers compare what they would have done versus what the machine recommended, revealing where brands are succeeding and failing – and providing valuable insights for the next season.

Glass box AI

Right now, “black box” AI applications produce results using algorithms with a level of complexity that only computers can understand. But computers don’t have the final say – humans do. And if those human decision-makers don’t understand it, they don’t trust it. While black box solutions serve their purpose, they also limit the value organisations can extrapolate by hiding AI logic, which in theory could be used to teach humans what was learned that led to various recommendations.

As AI adoption increases, we’ll see more organisations move to glass box AI, which exposes the connections that the technology makes between various data points. For instance, glass box AI not only tells you there is a new retail opportunity, it also uncovers how that opportunity was identified in the data. It also provides retailers with an opportunity to check their data – and any public or aggregate data they pull in – to ensure AI isn’t making bad assumptions under the adage “garbage in, garbage out.”

This may sound more complex, and it is. Even with next-gen UX that simplifies the integration of AI into processes and workflows, retailers must invest in educating employees to make the most of the predictions that AI delivers. But if we’ve learned anything in the past decade, it’s that data-driven insights aren’t a passing fad.

Trusting tech

While prediction-generating AI holds great promise, in order to fully realise its potential, retailers will have to overcome a few common challenges, most of which are found in underlying data. The burning question is, how much data is enough? And how do you trust it?

Sparse and intermittent historical information can quickly become a roadblock to providing useful AI-powered insights. With a lack of visibility into what’s happened in the past, retailers often fall into a cycle of using previous assumptions to predict what will be “en vogue,” which might not accurately represent their business from year to year. Operating without a holistic view into this information can lead retailers to analyse data that won’t yield productive insights, leaving them without a clear picture of their consumers’ demands. People are remarkably good at spotting patterns. However, to trust the technology, retailers need to look through the AI glass box and understand that these prediction tools are able to sift through volumes of data that humans can’t absorb and find patterns that people can’t see.

Ultimately, the customer data dilemma will impact every organisation looking to benefit from AI. The point is, there is a clear opportunity for AI to help retailers amplify their insights and improve their customer experiences.

AI is making waves in the retail industry by enabling prediction and improving merchandise planning. We’re also seeing development of glass box AI, which will allow humans to take a bigger role in data-driven decision-making. The fantastic thing about AI and retail is that people play a vital role in ensuring that we’re putting the right products in front of the right people, so the human touch will always be needed. Used in the right way, AI can continue to offer retailers major competitive advantages.

Nikki Baird, vice president of Retail Innovation, Aptos (opens in new tab)

Nikki Baird is the vice president of Retail Innovation at Aptos, a retail enterprise solution provider. She is charged with accelerating retailers’ ability to innovate.