All Articles

eMarketer flagged in early 2026 that retailer pushback could cloud the year's AI progress. The pushback is real, but the reasons behind it are more specific than AI fatigue. Three structural concerns, from retailers who understand the technology well enough to see the problem.

Adobe's March 2026 data shows AI-referred traffic converting 42% better than non-AI channels. That's a record, and a number that needs careful interpretation: selection effects, early adopter behaviour, and the specific nature of AI-mediated discovery all shape what it means for your planning.

TikTok Shop's 2025 full-year US GMV came in at $15.82 billion — up 108% year-on-year. The global figure was $64.3 billion. The AI recommendation engine driving that growth isn't a feature; it's the entire product. Brands that still haven't engaged with social commerce are running out of comfortable reasons not to.

Smart carts, autonomous checkout, computer vision for shelf intelligence, inventory robots. All have had conference moments. In early 2026, some are scaling. Some are quietly struggling to find their use case outside the pilot environment. The UK high street has a specific lens on this.

Digital Commerce 360 described early 2026 as a 'structural reckoning' for ecommerce. AI-powered shopping, agentic purchasing, tariff disruption, and zero-click search hitting simultaneously. The cumulative effect isn't incremental — it's a different operating environment. Here's how I'm thinking about it.

I spent an evening at The Lookout in Bishopsgate listening to Vercel's CPO, ElevenLabs, M&S, and Virgin Media O2 talk about building for the agent era. The view from the 50th floor was spectacular. The ideas were harder to shake.

Accio Work, launched by Alibaba International on 23 March 2026, is an enterprise AI agent platform built specifically for SMEs. No-code, multi-agent, and claiming to build an online store in 30 minutes. The adoption rate is real. What it actually means for smaller UK businesses is a more interesting question.

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.

Jeremy Howard proposed LLMs.txt in September 2024 as a way for websites to communicate with AI crawlers. By October 2025, 844,000 sites had implemented it. The major AI crawlers are still largely ignoring it. So where does that leave us?

AI-powered fraud in early 2026 isn't the high-volume, low-effort credential stuffing of five years ago. It's fewer attacks, smarter attacks, and attacks that are genuinely harder to distinguish from legitimate behaviour. The industry isn't losing — but it's not comfortable either.

Google AI Overviews now appear on 48% of searches, with an average zero-click rate of 83% when they do. Between 58% and 68% of all Google searches now end without a click. The organic traffic model that UK retailers have relied on for fifteen years is structurally broken. Here's how to think about that.

Only 39% of Americans trust AI agents to make everyday purchases on their behalf. That sounds like a problem for agentic commerce. Look more closely and it's more interesting than that: trust is real, category-dependent, and building along a predictable path.