Q&A: The AI honeymoon is over: The shift from AI experimentation to accountability
In retail, predictive analytics can shift teams from reacting to historical data to anticipating future needs. AI models track shopping frequencies and lifestyle changes (e.g., a sudden shift to organic or baby products) to personalize marketing in real time.
How can retailers harness such developments. Looking into this maturing technology and the associated strategies is Sean Turner, CTO and co-founder of Swiftly, who explains more in conversation with Digital Journal.
Digital Journal: In April, you said AI in grocery was becoming “less about experimentation and more about execution.” Has that shift accelerated in recent months, and what does execution need to look like for retailers today?
Sean Turner: Yes, but I think a lot of retailers are still trying to figure out what execution actually means. It’s one thing to give a team access to an AI tool. It’s another to connect AI to the data and systems it needs to actually understand the business.
If the only way AI can access your data is by having someone export a spreadsheet and upload it into a tool, you’re going to be pretty limited in what you can accomplish. In grocery, there’s so much more context available. You have every transaction, every customer interaction, every ad impression, pricing, promotions and loyalty behaviour. When AI can work across that information together, it can start helping retailers make much better decisions.
That’s where I think execution is headed. It’s less about experimenting with AI tools and more about giving AI the context it needs to solve real business problems and then measuring whether it actually worked.
DJ: What should grocery retailers be asking of AI now that the technology is moving from experimentation into a more accountable phase?
Turner: Retailers should be asking a pretty simple question: What is this actually doing for my business?
Grocery retailers are sitting on an incredible amount of data about how people shop. They know what customers buy, how often they shop, which promotions they respond to and how that behaviour changes over time. But most of that data has never really been put to work, because answering even a basic question across it can take an analyst days or weeks. AI changes that. For the first time, retailers can get answers about their business that they had no practical way to get before, and get them in minutes.
Take a retailer that just opened a new store. AI can tell them how many of those shoppers are new to the banner versus shifting over from an existing location, which stores they’re coming from and how much spend those stores lost as a result. A category manager can ask how produce performed last week and get talking points that already account for new stores inflating the numbers, flag which subcategories are slipping and call out the specific items driving the change. And a merchant can look back at every past promotion and see not just whether it moved the item, but whether it grew the whole basket, brought shoppers back the following week or pulled in new buyers. Most grocers have never had that level of insight into their own business.
And retailers should be able to measure the result. Did it lead to a better decision? Did it drive another trip or grow the basket? Did it show up in margins? That’s the level of accountability we should expect from AI.
DJ: Where are you seeing AI create the most meaningful value for grocers today, and where do you think expectations are still running ahead of reality?
Turner: The strongest value today we’re seeing today is in helping grocers make better use of the data they already have. AI is improving how retailers interpret shopper behavior and identify patterns at a level of precision that would be much harder to achieve manually. Personalization is one of the clearest examples. Instead of sending the same promotion to thousands of people and hoping it resonates, AI can help a retailer understand which shoppers are most likely to respond, what is relevant to them and whether that engagement actually changed what they purchased.
Where expectations get ahead of reality is when companies expect AI to create value without giving it enough context. AI isn’t going to magically understand your business. It needs access to the right data and systems. The richer that context is, the more useful AI becomes.
DJ: Grocery operates on notoriously thin margins. What does an AI use case need to deliver for a retailer to justify continued investment?
Turner: Grocery is a low-margin business, so the bar is pretty straightforward. AI has to help you make more money, save money or make your teams more effective. And you have to be able to prove it.
The biggest opportunity is hiding in the data retailers already have. A grocer can have millions of customers and tens of thousands of SKUs in the average store. That’s a massive dataset, and there are insights buried in it that nobody is looking at, like an item that’s quietly losing money in certain stores, a promotion that’s eroding margin instead of driving sales or a category where shrink keeps creeping up. Retailers can’t afford the analyst time to dig through all of that, so most of those insights never get surfaced.
That’s exactly what AI can do. It can work through that data continuously, find what matters and point the team to it. And when retailers see those insights and act on them, that’s how you grow margins and reduce waste in this business. That’s the return a retailer should expect from an AI investment.
DJ: As AI-powered shopping assistants and product discovery tools become more common, how could they change the relationship between retailers, brands and shoppers?
Turner: AI shopping assistants are going to change how people discover products, but how useful they are really comes down to how much context they have.
If I ask an AI assistant what I should make for dinner tonight, it can give me an answer pretty easily. But what if it also knows what I typically buy, what I bought last week, what’s on sale at my grocery store and what’s actually in stock? Now it can give me an answer that’s actually relevant to me.
That’s where retailers have a real advantage. They already have a tremendous amount of information about how their customers actually shop. The opportunity is to bring that intelligence into these AI experiences so they can go beyond generic recommendations and actually understand what’s relevant to that shopper.
DJ: Personalization has been a major promise of AI in retail. What separates personalization that genuinely improves the shopper experience from personalization that simply adds more noise and friction?
Turner: I think the personalization conversation in grocery has been too focused on the individual offer. Sending a shopper a slightly more relevant coupon is useful, but it’s not where the biggest opportunity is.
What actually improves the shopping experience is a store that has the right products, at the right price, on the shelf when people come in. And every store is different. What works in a suburban store full of families doesn’t look like what works downtown, and demand shifts week to week. Retailers have never had the ability to understand those differences at the store and item level, so a lot of decisions get made the same way across the whole chain.
That’s where AI cuts through the noise. It can surface what’s actually happening in each store, like an item that keeps going out of stock, a promotion that isn’t moving product or a category where shrink is creeping up, and recommend what to do about it. The shopper may never know AI was involved. They just find what they came for.
DJ: Looking ahead, what will distinguish the retailers that successfully use AI from those that invest heavily without seeing meaningful business impact?
Turner: Over the next few years, AI will compress the time between understanding shopper behaviour and a retailer taking action. Grocers have historically used data to understand what happened after the fact. You run a campaign, look at the results and use what you learned to make the next campaign better. AI can significantly accelerate that cycle. If demand for an item starts shifting in a group of stores, the retailer can spot that change much faster and adjust pricing, promotions or inventory while it still matters.
The retailers that get the most value from AI will be the ones that can act on those insights quickly. A shift in how an item or category is performing could inform the next promotion, a store’s assortment or where a retailer allocates media spend. The faster retailers can learn from what’s happening and apply that intelligence to the next decision, the more valuable AI becomes.
Q&A: The AI honeymoon is over: The shift from AI experimentation to accountability
#honeymoon #shift #experimentation #accountability
