Generation Next Thinking: How GenAI can reshape the way grocers do business
Generative AI is generating a lot of buzz, but it’s not just another shiny new object, according to experts. “I think we’re on the precipice of the biggest industrial revolution that we’ve seen in our lifetime,” says Lindsey Mazza, global retail lead at Capgemini Group.
Indeed, generative AI (or GenAI)—a type of artificial intelligence that can quickly generate content, have human-like conversations and produce insights from large volumes of data—is being touted for its transformative powers in many sectors, including retail. With the integration of AI and natural language processing, experts say GenAI has the potential to change how people work, how businesses operate and how they serve customers. And it’s happening fast.
ChatGPT—a consumer-facing, GenAI tool launched by tech firm OpenAI in November 2022—reached 100 million active users just two months after launch. (For context, it took Instagram two-and-a-half years to reach 100 million users, according to data from Sensor Tower, as Reuters and other media outlets have reported.) “When we see that kind of progression and escalation of the scale and rapid user acceptance from the end consumer, we have an expectation that organizations will then also use these kinds of technologies to support their business,” says Mazza.
READ: What ChatGPT and AI could do for the food industry
They’re certainly talking about it. In a recent global study of industry leaders by Capgemini Research Institute, Harnessing the Value of Generative AI, nearly all respondents (96%) cite GenAI as a hot topic of discussion in their respective boardrooms. In retail, 59% of executives say their organizations’ leaders are strong advocates of GenAI, while 37% are taking a wait-and-see approach. Meanwhile, a recent study by Accenture found 97% of senior executives agree that GenAI foundation models (also known as large language models or LLMs) will be transformative for their company.
How big is the potential impact? McKinsey & Co. estimates that GenAI’s impact on productivity could add up to US$4.4 trillion in value to the global economy annually, with roughly 75% of that value coming from four areas: customer operations, marketing and sales, software engineering and R&D. In retail and consumer packaged goods, the potential impact of GenAI on revenues is significant, at US$400 billion to US$660 billion a year.
Gary Hawkins, chief executive officer and founder of the Center for Advancing Retail & Technology (CART), believes there is enormous potential for this emerging technology. “AI and generative AI, in particular, is absolutely transformative when it comes to retail,” he says. “We are still early on in this whole evolution. I don’t think a lot of people in the retail industry truly understand what AI and GenAI is and how it can be used to transform their business… But it’s moving.”
So, what can GenAI do for grocers? In Hawkins’ view, one of the main advantages of GenAI is its ability to automate business processes. “What that automation brings to the retailer is not only cost reduction in the form of needing fewer human beings to do [certain tasks], but it typically results in better decisions, much faster cycle times and more effective output,” he says.
In Capgemini Research Institute’s study, executives agree the benefits of GenAI are “not only in consumer-facing and experiential areas, but in operational ways as well,” says Mazza. For example, 78% of executives believe GenAI will make product and service design more efficient, 71% believe it will help them improve the customer experience, and 65% say it will help improve internal operations.
Here’s a look at a few ways GenAI can come to life in grocery retail.
1. Content
GenAI is the talk of the marketing town, as it can be used to write content such as ad copy, social media posts and product descriptions—all in a brand’s own voice if it’s trained on it. “The main benefit of generative AI at this point is to be an accelerator or a productivity enhancer, so for a product description, it can give you a good first draft to work from, especially when you can plug in information to make the language sound like the brand,” says John Harmon, managing director of technology research at Coresight Research.
GenAI can also be used to create visuals using simple language prompts. Technology company Nvidia has a GenAI solution called Picasso that lets businesses create images, videos and 3D content, training the models with their own data and assets. “In literally seconds you can talk to it and say, ‘show me a bottle of [insert brand here] with a cheeseburger at a campfire during sunset’ and this is what it creates,” says Azita Martin, vice-president and general manager of AI for retail, CPG and QSR, at Nvidia. “Consumer packaged goods companies and grocery companies that do a lot of their media and marketing in-house can leverage this type of multimodal, generative AI model to very quickly create visually stunning campaigns that they can use for email campaigns, for advertisements, for social media campaigns and so forth.”
2. Personalization
There’s a big personalization play with GenAI: It can analyze large amounts of customer data (such as purchase patterns and demographics) and generate targeted, personalized content almost instantly. “In the area of marketing personalization, the leading-edge solutions I’m seeing use both AI and generative AI to very efficiently understand what specific promotions and offers to make for each individual customer to get whatever the desired result is—more shopping frequency, bigger baskets, improved retention—and be able to do this at scale incredibly quickly,” explains Hawkins.
READ: How embracing artificial intelligence can elevate the shopper experience
3. Search
For consumers, GenAI can level up the online search experience. Earlier this year, Walmart in the United States launched a GenAI search feature that lets customers key in an event or activity, rather than searching for specific products or brand names. For example, instead of typing “cheese snacks” or “football décor,” a customer can search for “football viewing party” and get personalized product recommendations.
“Even though we’ve come a long way with our search engines… if you’re searching for a themed birthday party, you would get [results based on] the terms, not necessarily suggestions for making your birthday party run better,” says Michon Williams, chief technology officer at Walmart Canada, which is working on bringing the new search tool here. “GenAI creates a whole other level of sophistication in the responses because it has more context … which means you can make better recommendations.” What’s more, Williams says the recommendations are not just based on search, but also on the context of the customer’s shopping habits and patterns, which makes the recommendations even more relevant.
READ: Inside Walmart Canada’s tech transformation
4. Employee assistants
Last year, Walmart rolled out My Assistant, an internal GenAI-powered chatbot for corporate head office staff. Walmart says the tool can do everything from speeding up the drafting process, to summarizing large documents or meeting notes, to serving as a collaborative partner. “It helps free up time from daily tasks and enables us to focus more on the bigger, strategic things that require big thinking and creative solutions,” says Williams.
For example, My Assistant “can take summaries of a meeting and call out what the next best actions could be for us to take,” says Williams. On the more creative side, the tool can help associates plan things such as in-store features. For Walmart Canada’s recent 30th anniversary, for example, Williams says a home-office associate could ask for ideas on how to celebrate 30 years and to recommend foods from the ‘90s.
5. Shopping assistants
Grocery shoppers can also have their very own GenAI assistants at their fingertips. Capgemini recently launched “Casey,” a “conversational commerce” digital assistant that helps customers build customized meal plans and find recommended products and ingredients in stores. Casey can also provide tips on topics such as nutrition, mental wellness and immune issues.
The technology allows grocers to “engage with the consumer in a new way and create loyalty” by giving them a more convenient in-store experience, says Mazza. “It also fulfils some needs in their life of planning those meals, which is part of the mental load of going to the grocery store, doing the shopping and thinking about what they’re going to cook when they leave the store with all of those items.” All this ladders up to bigger baskets, topline sales growth and improved margins, according to Mazza.
6. Data and analytics
For Coresight’s Harmon, one exciting area of GenAI is the ability to integrate a company’s own data into a large language model (through a technique called retrieval-augmented generation or RAG). “I call it talking to your data—you can ask questions about your own company’s data and generative AI will dig it up,” Harmon says. “This democratizes the use of generative AI, so you don’t need to have a PhD or be a data scientist. The prompts are written in natural human language.” For example, a person can ask, “what was our sales growth for the last three years?” or “what are opportunities to increase revenues?” explains Harmon, and the tool generates answers and ideas.
READ: Sklar Wilton’s Michael D’Abramo on how to handle data
On the supply chain side, the integration of company data can bring significant value. “[A use-case] we’re seeing is around asking supply chain questions like ‘how much inventory is there in the distribution centre in my region?’ and ‘how quickly can I get this product replenished?’ and so forth,” says Nvidia’s Martin. This can help retailers avoid stockouts and overstocking, as well as help with demand forecasting.
Pitfalls, perils and best practices
While GenAI has clear advantages, there are many considerations grocers must address before going full swing. First and foremost is data, since the quality of GenAI outputs is largely dependent on the quality of data it’s trained on.
“The quality and cleanliness of a retailer’s data is becoming truly mission critical,” says CART’s Hawkins. “AI and GenAI feed off data and if their data is not good—if it’s corrupt or there are problems—the output is not going to be good. And so, retailers need to finally start paying attention to the quality of their data.” He also points out that larger retailers are building their own language models rather than using public systems such as ChatGPT. “Making it truly specific to their retail organization helps ensure that the output is accurate relative to their enterprise,” he says.
Nvidia’s Martin reiterates the popular “garbage in, garbage out” adage about data. “With generative AI, the more [quality] data you give it and the more data you train with it, the better the accuracy,” she says. “And so, having the best data and clean data is the bare minimum of getting accurate models.”
Retailers need to be aware of the risk of drift—meaning the system’s performance and behaviour changes over time, drifting away from the training data in unpredictable ways. As Coresight’s Harmon puts it, language models are “living organisms” that change over time. “All AI models drift because as you use them, they try to get better and better and they change their ability to generate results,” Harmon explains. “So, for all AI models, including generative AI, you have to monitor them for drift, accuracy biases and toxicity.” That means organizations must bring governance—including policies and procedures—to AI models.
Capgemini’s Mazza suggests retailers treat GenAI like an intern. “They’re very smart… but need a lot of oversight,” she says. “We have to provide the context and evaluate that we’re being good digital citizens with the data that we have access to. And so, it requires the oversight, governance and organizational structure to support it.”
READ: Ottawa puts up $50M in federal budget to hedge against job-stealing AI
That’s certainly a focus at Walmart Canada. “We have strong controls around privacy and consent, but internally, there’s more work to do as the volume of data increases,” says Williams. “So, a big challenge is continuing to have the right people with the right experience to massage that information and [ensure] it can provide meaningful recommendations because it’s well managed and well defined.”
One thing is clear about the future of GenAI: It’s moving fast. “It’s going to do nothing but accelerate,” says Hawkins. “Anyone in the retail industry needs to understand we’re now living in a world that has never existed before. And anyone thinking that change is going to happen at the same pace it did last year or last week is fooling themselves. It’s only going to happen faster and faster.” That means retail leaders who are leveraging AI are only going to widen the performance gap versus slower-moving retailers, he adds.
Harmon thinks back to when ChatGPT burst on the scene and people used it to write poetry and essays. “Now, we’re hearing about using generative AI to bring real-traffic data into route planning … We’re hearing about using generative AI in store design, optimizing supply chains and managing supply chains. This technology is moving fast,” he says. “And the best part is there’s no programming involved. You just ask it what you want, and it does it for you.”
Generation Next Thinking is an ongoing series that explores the cutting-edge topics that are impacting grocery retail today and in the future.
This article first appeared in Canadian Grocer’s May 2024 issue.
