In this episode of Marketplace Masters, Paul Sonneveld welcomes back Chris Turton, Managing Director of Ecommerce Intelligence in the UK, for a candid revisit of a conversation they had eighteen months ago, when Chris was distinctly unconvinced that AI was moving the needle for agencies.

The fear in the market is familiar: AI makes execution faster and cheaper, clients expect fees to come down, and agency margins get squeezed. Chris thinks that’s the wrong hypothesis. For his agency, AI has been about creating client value, not cutting cost. He explains what’s actually changed in his thinking (AI shopping and agentic commerce, yes; generative engine optimisation on Amazon, still not convinced without better data), and what he’s genuinely automated, including a Claude-built system that handles a 5,000-variation vendor catalogue problem that used to consume hundreds of hours.

Paul and Chris also get into the trap of treating AI output as “done” when every listing on page one scores ten out of ten, whether clients are asking for AI discounts at renewal, how freed-up time gets reinvested without being burned on peripheral projects, how agency team structures are shifting toward performance and numbers-led roles, and why multi-retailer visibility is becoming the thing that makes you better on Amazon.

Chris closes with the mistake he’s made repeatedly: building AI tooling without starting from the end in mind.