Generative AI 

Generative AI has been a significant driver of much of the recent hype around the impact of AI technologies. But, while the algorithms behind these services have clearly become more advanced, it’s not necessarily the complex, behind-the-scenes tech that has led to the recent leap in awareness and interest.

Instead, says Ocado’s Jensen, its success is related to how easy it is to use. “The single most important trigger behind AI in the news cycle is not even the tech behind apps like ChatGPT. It was the addition of a user-friendly interface to that tech, making it truly accessible to anyone.

“Suddenly, billions of people with no technology background have access to a toolkit that until now was solely the domain of machine-learning engineers. In retail terms, it’s the equivalent of adding a shop window.” 

Generative AI use is already widespread. A global McKinsey & Company survey of 1,684 employees in April 2023 found that 40% of respondents working in the consumer goods and retail industries have tried it at least once, while a further 18% regularly use it in their work.

As with other industries, the most popular uses so far include marketing, product development and customer care.  

Generative AI’s potential benefits include faster product development, enhanced customer experience and improved productivity. Its uses can even stretch to supporting store design and developing more creative ways to get closer to customers by inspiring new customer engagement strategies.

And its long-term impact shouldn’t be underestimated. Gartner predicts generative AI will have “an impact similar to that of the steam engine, electricity and the internet”.

Several global retailers, including ecommerce behemoth Amazon and fashion retailer Zalando, have implemented generative AI throughout their businesses. The former has invested heavily in large language models (LLMs) and generative AI, with the end goal to transform and improve “virtually every customer experience”.

Amazon also revealed in August 2023 that it would be using generative AI to produce summary paragraphs of product review sentiments and key themes, to help customers determine at a glance whether a product is right for them. In the testing phase at the time of writing, this use is set to be rolled out more broadly over the coming months if successful.  

Zalando, meanwhile, rolled out AI-based, virtual fitting-room technology across all 25 of its markets in April 2023, with use of the fit tool becoming an expected part of the online shopping experience for its customers. Stacia Carr, vice-president of size and fit at Zalando, said: “We can already see that customer engagement with these campaigns increases and, in fact, around half of the customers try more than one size on the avatar”, showing clear benefits of the tool for its shoppers.

Customers create a 3D avatar by entering their height, weight and gender, and can then use this to see how different sizes from various brands would fit them. A combination of fitting models, machine learning, computer vision and other technologies predict if items will run big or small.

As usage of Zalando’s fit tool demonstrates, while the hype will die down as the intricacies and reality of AI implementation set in, we can expect increasingly innovative applications as people and businesses discover more ways to use it.  

Getting the right tone 

At the moment, generative AI does have some limits, showing best success when paired with individuals and used as the starting block for content outputs. Alone, its outputs run the risk of inaccuracy and bias, depending on the material it is fed to learn from.

With one of the biggest uses in retail so far being the creation of marketing or product content, can it be trained yet to produce content in the right tone and using the right brand guidelines? Is it possible for AI to get the subtleties of branding just right, applying the right level of humour, emotion and personality for any particular brand?

As the experts in this report discuss, it seems an integrated approach, using AI to bolster the work of retail colleagues, is currently the best one to take.

PwC’s Byrne says standard generative AI platforms that are publicly available, such as ChatGPT, are unlikely to be able to echo brand messaging accurately because they are not trained using the right inputs.

“They might be an efficient way to get started for idea generation, but you need to use your own data or build your own models to get your messaging, tone of voice, values and ways of thinking to create something that’s personalised to your brand,” she says.  

None of this is easy and the right skills can be hard to recruit. As a result, third parties are playing a significant role in helping retailers to ease their way through.

The extent to which they help all depends on the use case, says Matt Moorut, director and analyst at insights provider Gartner’s Marketing Practice.

“For simple tasks such as building out copy for product page descriptions, third parties aren’t really needed, since marketers can use freely available tools to streamline their work,” he says. “For more advanced use cases like product design, or where core operations need to be adjusted, then third parties make more sense.”  

The Very Group  

The Very Group has long been a leader when it comes to technology investment. Its strategy is to create long-term relationships with customers via personalised experiences and it has invested accordingly over the past decade. In 2015, it claimed to be the first in the sector to launch a fully personalised homepage.

The retailer started investing in AI in 2017 when it introduced its Very Assistant chatbot. Very now uses AI for a variety of tasks, including fraud prevention, deciding the ideal frequency of emails to send to each customer and dynamic pricing.  

In recent years, the company has launched two relevant business-wide programmes.

In summer 2021, it started a three-year project to transform the technology and data that powers the business, with a goal to migrate its app and website to its new Skyscape ecommerce platform by 2024. The hope is that this platform will enable it to deliver customer experience changes more quickly and more frequently, through one dynamic content management platform.

In March 2022, the group announced plans to transform the technology behind its ecommerce platform through third-party partnership.  

Very’s personalisation efforts are now driven by AI and it continues to work on improving areas such as marketing emails, the relevancy of ads and increasingly personalised homepage content. It has also trialled real-time and customer-specific offers via push notifications in its app, which generates 45% of its total sales.

Clothing shape and fit can be a barrier for online fashion shoppers and Very is using AI to tackle this as well, partnering with True Fit in May 2022 to provide personalised guidance across its 300 fashion brands. True Fit uses AI-powered machine learning to connect shoppers with items that fit, with the aim of reducing returns.  

Very is also aiming to improve how customers discover new products on the site, using AI to learn from customer interactions and make better recommendations for products.

In its first phase, tech is being used to improve search and auto-suggest tools on the site; in the longer term, the retailer will roll out other functionality, such as a quiz that will help customers find the right products.

Speaking to Retail Week in June 2023, Very’s digital customer experience director Paul Hornby said the tech “will make it so much easier for customers to find the product they love”.  

Other AI-related services from Very include in-app augmented reality (AR) that allows customers to try out new cosmetics products from L’Oréal, Maybelline and NYX digitally. It uses ModiFace AR technology, which relies on face, hair and nail tracking, analysis and rendering, and was launched in December 2022.  

Ocado  

Ocado Group, which is made up of tech leader Ocado Solutions and grocer Ocado Retail, is another company leading the way with AI investment.

AI is used throughout the customer journey, primarily via personalisation – customers’ favourites and previously purchased items appear at the top of searches, and the company sends reminders for items they purchase regularly, such as their weekly milk.  

Outside of direct customer engagement, another major focus in the grocery arm is reducing waste. The company says its AI-enabled forecasting leads to waste levels that are 50% lower than industry norms, achieving this via 20 million demand forecasts per day for what customers will order and when.

This forecasting technology also predicts future stock levels, leading to higher availability and lower substitutions for customers.  

Ocado Solutions has been developing AI-powered robots to pick and pack groceries since 2015, which it offers to third parties and uses across its Ocado Retail warehouses.

Using an air-traffic-control-style system, items are picked in the best order for speed and accuracy. The company says a 50-item order can be picked in five minutes, compared with more than an hour in a physical store.

In 2022, Ocado announced its Orbit project, which splits an order across a number of smaller fulfilment centres, creating a “virtual distribution centre” powered by AI and machine learning. The service means a wider range can be offered with a shorter lead time to a higher number of UK households.  

Shiny object problem  

One big question for retailers is where to invest. As discussed in previous chapters, AI is impacting every part of the value chain and has the potential to improve operational efficiencies and customer engagement across the board. So where should retailers set their priorities?  

The answers differ significantly according to the business. For example, the likes of the Co-op are rolling out AI-powered robot deliveries, which use a mixture of cameras, GPS and sensor fusion to navigate. Meanwhile, brands such as L’Oréal are using AI to launch virtual try-on services to help customers make decisions when shopping online.

But do customers actually want these services? And how can retailers determine where to put their resources?  

Beauty retailer Space NK brought this question to the fore when developing its store strategies. The retailer, which largely targets affluent shoppers with its premium range of beauty and wellbeing products, is limiting the impact of AI in stores, based on customer feedback from its currently successful model.

The business saw its revenues rise to “record-breaking” levels in the year ending March 2022 with sales of £119m, up 6.5% on the previous year.  

Speaking to Retail Week in May 2023, Space NK chief executive Andy Lightfoot emphasised the importance of understanding customer wants and needs when creating engaging retail spaces.

“Experiential retail is one of these overused terms that could be engineered to mean absolutely anything,” he says. “For me, the experience that customers want when they go to a beauty retailer is to be able to touch the product, use it, play with it, feel it, smell it and have a really positive interaction with a member of staff should they wish.  

“So you won’t see loads of screens in our stores. You won’t see loads of artificial reality experiences. We know that isn’t why our customers go into stores; they can do that on their phones.” 

Following a £100m AI tie-up with Google, announced in August 2023, John Lewis Partnership (JLP) considered the potential of AI-driven, Just Walk Out-style checkouts – like those in Amazon Fresh stores – at Waitrose and how innovating this part of the shopping experience may not be right for the brand and its customers.

JLP chief technology officer Zak Mian said: “The jury’s out if customers want assisted service in a grocery shop or for it to be fully automated. A lot of our customers like our service of the fish, meat and cheese counter at the back of the shop – that makes a difference.” 

Brendan Witcher, vice-president and principal analyst at research firm Forrester, says the key to making AI applications a success is to ask whether they solve an actual problem for customers. AI-powered drone deliveries, for example – being trialled by Tesco in the UK, Walmart in the US and Coles in Australia – is one interesting idea to support customer convenience.

However, Witcher adds: “I don’t know a single customer who stands at their front door and says: ‘That package better come by drone or I’m never ordering from them again.’ You don’t really need it – that’s the shiny object problem.

“It’s interesting and it’s sexy, but are you really going to increase sales because of it? Are customers only going to order from drone delivery companies? We don’t care how the package gets to us.”  

By contrast, virtual try-on services such as Zalando’s could help to tackle a real problem customers face when shopping online: knowing what something will look like on them. “That elevates the customer’s confidence in buying and AI is going to be most impactful when it removes pain points from the customer experience,” says Witcher.

The goal needs to be prioritising technologies that either reduce costs or increase revenue, he adds. The environmental and cost implications of returns also cannot be ignored here, with increasingly accurate virtual try-on solutions reducing another headache for retailers.