If you run a retail business with 40 or more stores, you have probably been told AI will transform your operations. The harder question is how to roll it out so store teams use it and the results show up in your numbers. Global brands are already looking beyond implementation to how they build resilient operations that deliver lasting value from their AI investment.
This October, at FT’s Future of Retail 2026 event, we caught up with leaders from some of the biggest retail brands in the world to dive deep on future proofing margins with the help of AI.
Zipline’s Chief Product Officer, Meagan Sobal, spoke alongside leaders from Lululemon, Kingfisher, and Adobe on a panel, titled ‘Operational efficiency — how retailers can use AI to protect margins’. Here are the five lessons from that conversation that you can apply right away when you plan how to implement AI in your own stores. In short: lasting AI adoption in retail depends on frontline trust, a clearly measured outcome, visible time saved in store, transparency at headquarters, and strong operational fundamentals.
Lesson 1: Rolling out AI
You can’t just plug a large language model into a retail operation and expect instant results. Having seen 70+ brands roll out AI-assisted chat experiences, Sobal says the starting point isn’t onboarding a new tool: it’s building trust with the people who use it.
“From our experience, […] implementing an AI to improve a workflow or to give time back to your frontline teams is actually incredibly hard. It’s not something that you can just plug an LLM into and be like, alright roll this out,” says Sobal.
“There’s a lot of trust that needs to be built not only with the people using the technology at the end of the day but also in governance and security and evaluations and testing in order to make sure that you’re handing your frontline teams an answer that they can rely on.”
Key takeaway: Adoption is the real measure of whether an AI tool is working. If you lose the trust of your frontline teams, you lose adoption, and without adoption the technology is just another solution. Before you scale, put governance, security, evaluation, and testing in place so they get answers they can rely on.
Lesson 2: Start with the outcome, then measure it
Before choosing a tool, define the outcome you want to improve and establish how you measure it today. Then look at the workflows that contribute to that outcome and test whether changing them has a positive or negative effect. That’s how retailers separate correlation from causation, working out what actually moves the needle.
According to Sobol, the conversation frequently isn’t about labor efficiency or the bottom line, but about what outcome is worth improving. Measurement is where many AI pilots fall short.
“You’d be surprised at the number of pilots that I’ve seen that really haven’t had the right measurement framework and then you’re not really understanding properly where to make your investments, what’s working and what’s not,” explains Adobe’s Gemma Carver.
Defining the outcome also tells you where to start. Former Lululemon CIO, Julie Averill, highlighted the tension between knowing that AI can support the business and being unsure where to begin. Her suggestion? Look at the operational side of things.
“I think the reality of it where we’re really seeing returns is in the operational side of it where there’s real problems, there’s real inefficiencies to go after. You start small, you build something, and then you can deploy broader,” Averill explains.
In practice, that means picking one operational process with a clear inefficiency, recording your baseline, testing the change in a small group of stores, and expanding only once the numbers move.
Zipline CEO Melissa Wong wrote about a similar trend in Forbes, where the need for AI is presumed, but the questions retailers are asking vendors and the attitude of what they’re asking AI to do for them are pointed at the wrong priorities.
Key takeaway: Know the problem you’re solving for. Don’t make the assumption that AI technology will transform your organization through speed or volume of output. Start with your actual organizational priorities and build the technology around the outcomes.
Lesson 3: Show the value and give time back to the frontline
Store teams don’t fear technology itself. They want to do a good job and not be told the wrong thing. Optimization looks different at HQ than it does on the frontline, so the value of any new tool has to be visible where the work happens.
Sobal explains that: “It’s really about proving the value of the tool that you’re providing to your employees and bringing them along on the journey, explaining the why with the what, helping them enroll behind the reason why this technology is being offered to them to assist them in their day to day and then actually deliver on it.”
That’s the principle in practice. In a pilot with a large UK retailer, a time study found that Zippy, Zipline’s conversational AI assistant, saved store managers about an hour a week by removing trips to the back room to find where a product lives. Accurate answers and visible time savings are what turn skeptics into users.
Key takeaway: Explain the why behind the what. Look for opportunities to remove friction from frontline workflows and give employees time back.
Lesson 4: Be transparent at headquarters, too
Trust has to flow from headquarters to in store teams. Leaders approving an AI investment need to see how a result was produced, not only the headline number. Sobol describes how Zipline handles this with its co-innovation partners:
“We walk our co-innovation partners through every single data analysis. We open the curtain up, we show them where […] the correlation is between what we’re seeing and the outcomes that they could potentially get and that just allows for a really transparent conversation.”
In practice, that means walking partners through every data analysis and showing how the correlations connect to potential outcomes. The aim is a conversation where AI isn’t a black box, and retailers understand the mechanics of their business and keep control over the results. It also builds on the second lesson: a transparent analysis is how you tell correlation from causation before you commit budget.
Before you sign off on an AI pilot, ask the vendor to show you the data behind the analysis, how the correlations connect to the outcome you care about, how impact will be measured, and how you keep control of the results.
Key takeaway: Transparency builds the trust that turns pilots into scaled programs. If a partner can’t show how it reached a result, treat the claim as unproven.
Lesson 5: Not everything has to be AI
AI is well suited to jobs like connecting data too vast for humans to interpret and linking it with human behavior. But Sobol is clear that it isn’t a replacement for good operations.
“There’s still so much hype from just the notion of AI and what AI can do […] more like a magic wand instead of really working through the hard parts, like data and people,” says Julie.
Meagan agrees. “There is still value in the way that you run your business. There’s still value in communication. There’s still value in making sure your teams understand what they’re supposed to do and when and have channels to give you feedback.”
Key takeaway: Apply AI where it adds real value and keep investing in the fundamentals. Used where it fits, AI can optimize existing processes and create new value.
Building a stronger foundation for retail AI
Most of the advice from this panel comes from instincts the best organizations already have. Build frontline trust before you scale, define and baseline the outcome you want to improve, show store teams the value and give them time back, stay transparent with HQ and partners, and keep investing in the fundamentals of good retail operations.
At Zipline, we’re constantly working with our customers to co-create AI products that make a difference for HQ and frontline teams. Keep up with the latest on AI-Powered Message Summaries, Zippy’s MCP capabilities, and, who knows, maybe we’ll see you at Future of Retail 2027 to go even deeper on how AI is shaping the future of retail operations.
