AI Personalization

LLMs are genuinely changing how recommendation engines work. UK shoppers are using AI tools in growing numbers. Most of them can't name a single experience that impressed them. That gap is the story.

After years of underwhelming chatbot experiences, AI-powered conversational commerce is finally delivering on its promise. Here's what's changed, and what the numbers actually say.

Tilt launched Snap in April 2026, an AI feature that creates product listings from live video in under a second. Early testing shows a 47% uplift in sell-through rate. The mechanism is less glamorous than it sounds, and more interesting than the headline suggests.

AI-powered loyalty programmes in 2026 can predict churn 60 days out, personalise incentives at the individual level, and adjust offers in real-time based on behavioural signals. Marks & Spencer, Tesco and Boots are already doing versions of this. The results are meaningful. The questions about data and consent are overdue.

True one-to-one personalisation (not segments, not 'customers like you') is in production at scale in early 2026. The infrastructure is real. The results are real. So are the questions about data, consent, and where helpful ends and uncomfortable begins.

The gap between virtual try-on demos and virtual try-on that consumers actually use has been wide for a long time. In 2025, it narrowed meaningfully. The question now is whether it closes all the way — and what retailers should be doing about it.

Amazon's new AI shopping assistant is imperfect and occasionally baffling. It's also probably the most commercially significant thing to happen to product discovery in years. Not because of what it does now, but because of what it implies about product content strategy.

When Klarna revealed its AI assistant had handled 2.3 million customer conversations in a single month, the industry took notice. The story behind the numbers is worth examining carefully.