五月 30, 2025
Despite the explosion of interest in generative AI, most retailers are still stuck in pilot mode.
At our recent retail roundtable with Google Cloud, we learned nearly 85% of retailers are experimenting with AI. But only 15–20% have scaled it in a meaningful way.
The blockers? They’re not technical. They’re about people, process and ownership.
Why do AI pilots stall?
The tech is ready. The business often isn’t.
- said Matt Hildon, Retail Portfolio Director at Valtech
Panelists across the room echoed that sentiment. It’s not about a lack of ambition but about readiness. Lewis Sullivan, Head of Strategy and Programs at easyJet holidays, shared how his company’s move from pilot to production wasn’t one big leap. It was a series of small wins that compounded over time.
Even with a strong AI stack and promising use cases, progress often halts after proof of concept for any of these reasons:
-
No clear owner. Pilots often start in tech or data teams but lack a business lead who can own the value.
-
Lack of repeatability. Successful pilots aren’t designed to scale.
-
Unclear metrics. Without useful feedback loops, there’s no case to invest further.
What scaling an AI pilot looks like in practice
Matt shared how several brands are getting past these blockers:
-
At Matalan, generative AI was deployed to help content teams. Product descriptions that once took hours were generated in minutes, freeing up creative energy and improving retention.
-
At Not On The High Street, Google’s AI-powered search tools drove a clear uplift in conversion.
-
At CPC Farnell, smarter search worked so well that suppliers began calling to ask what had changed.
Each case solved a real business problem. And each one showed impact quickly.
From experiments to execution
Moving AI from pilot to production means putting the right systems in place. These systems need to support:
-
Rapid validation and prioritization of use cases
-
Clear, consistent ways to measure business impact
-
Organizational alignment and repeatable deployment
At Valtech, we call this the catapult process, a lean loop that helps brands validate quickly, scale what works and leave behind what doesn’t.
We’re two years past the experimentation phase. AI is already delivering value. The next step is building the ownership, measurement and execution muscles that turn early wins into enterprise-wide impact.
Want to see how this plays out in the real world?
Explore how brands like Matalan and Not on the High Street are scaling AI with Valtech and Google Cloud. Read the case studies here.
