50 Million Shopping Visits a Month: UK Shoppers Are Ahead of UK Retailers

The Demand Side, in Numbers

Metapack's Ecommerce Delivery Benchmark Report 2026, produced with Retail Economics, put a number on something most retailers had only sensed: chat platforms such as ChatGPT now take 50.2 million shopping-intent visits a month in the UK. That places conversational AI alongside the country's larger ecommerce destinations as a place where buying decisions begin.

It is worth knowing how solid that base is. The research surveyed more than 8,000 nationally representative consumers alongside 400 senior decision makers at ecommerce businesses turning over $12.5m or more, across eight markets including the UK, in November 2025. This is not a vendor poll of a few hundred self-selected respondents.

The attitudes behind that traffic are the more interesting finding. 64% of UK consumers express trust in AI-assisted ecommerce, the highest confidence in Europe, and 30% of UK adults are open to an AI acting as a personal shopping agent: recommending products, checking delivery and returns, and making certain purchases once preferences are set. Among UK adults under 45, almost half already use AI tools for product research, price comparison and working out delivery options.

"AI is changing the way people find, choose and buy products, as well as how they are delivered. Retailers who adopt AI now will get ahead," said Al Ko, CEO at Auctane, Metapack's parent company.

Separate work by Worldpay, cited in the Payments Association's analysis of UK agentic commerce, points the same way and adds a generational split: 31% of UK shoppers would let an AI agent shop for them, rising to 45% among 18 to 34 year olds. Those shoppers expect agents to handle around 7% of their online purchases by 2030, roughly £29 billion a year.

Why 7% Is Bigger Than It Sounds

Seven per cent by 2030 is a modest-sounding forecast, and it is worth being honest that forecasts of this kind are frequently wrong in both directions. But three things make it more consequential than the headline suggests.

It is concentrated. A 45% willingness among under-35s against 31% overall means the effect is not spread evenly across your customer base. For a brand skewing young, the near-term number is not 7%.

Influence runs far ahead of transactions. The 50.2 million monthly visits are not 50.2 million purchases; they are the research phase. A shopper who forms a shortlist inside a conversation with an AI and then buys on your site directly counts as direct traffic in your analytics. The channel that shaped the decision is invisible in the report that follows it.

The winner-takes-most dynamic is harsher. Being on page one of a search result puts you among ten options. Being absent from an agent's answer means being absent entirely, because the agent gives one recommendation rather than a list. A 7% channel where a handful of products take nearly all of it is not a 7% opportunity spread thinly.

The Supply Side Has Not Kept Up

Intent is not the problem: 80% of UK retailers expect online sales growth in 2026, and the same research finds retailers worldwide planning to increase AI spending over the next 12 to 24 months. Readiness is another matter. The Payments Association's Q1 2026 survey of 100 senior finance, payments and risk leaders at UK online retailers found 58% believe AI-initiated transactions have already reached their platforms, and that only 41% feel very confident in the liability frameworks they are already operating under.

"In 2026, the focus shifts from experimentation to execution, where success will be shaped by how effectively retailers can embed AI into their data, systems and everyday operations," said Richard Lim, CEO of Retail Economics.

Into their data. That is where the gap actually sits, and it is why buying an AI tool does not close it. The obstacle for most retailers is not strategy or appetite. It is that their product data was written for people: a description that reads well, an image that sells, specifications in a PDF, stock levels that live in an ERP and reach the website twice a day. None of that is legible to an agent trying to decide whether your product solves a specific stated problem.

What Closes the Gap

The work is unglamorous and mostly consists of stating things a shop assistant would say out loud:

• Say what the product is for. Not its category and features, but the problems it solves and the situations it suits, in the shopper's language rather than the catalogue's.
• Answer the pre-purchase questions in the data. The ten things buyers ask before choosing, each with a short factual answer, expressed so a machine can read them rather than buried in a reviews tab.
• Keep price and stock live. An agent that recommends something unavailable damages its own credibility, and learns to stop using the source that embarrassed it.
• Be discoverable where agents look. A structured feed the major systems can find and fetch, rather than hoping a crawler infers the right thing from a rendered page.

Smaller retailers have a genuine advantage here, which is not something one gets to say often. There is no five-year platform roadmap to renegotiate and no committee to convince. A well-structured catalogue of 300 products can be more useful to an agent than a badly structured one of 300,000.

Where Agenticfeed Fits

Agenticfeed exists to do exactly this conversion: it takes an existing catalogue, from Shopify or a Google Merchant Center feed, and produces the structured, intent-rich version agents need, with each product mapped to the questions, problems and use cases it addresses, with grounded answers and live price and stock, published where the major AI systems can reach it.

If you want the mechanics of how an agent decides between two similar products, How AI Shopping Agents Choose Products covers it. If the immediate question is what any of this looks like in practice, start with What Is an Agentic Feed.

Frequently Asked Questions

How many UK shoppers actually use AI to shop?

Metapack's Ecommerce Delivery Benchmark Report 2026, produced with Retail Economics from a survey of more than 8,000 consumers across eight markets, put UK shopping-intent visits to chat platforms such as ChatGPT at 50.2 million a month, with 30% of UK adults open to an AI acting as a personal shopping agent and 64% expressing trust in AI-assisted ecommerce, the highest confidence in Europe. Almost half of UK adults under 45 already use AI tools for product research and price comparison.

Is agentic shopping actually going to be significant, or is it hype?

Worldpay research suggests UK consumers expect AI agents to handle around 7% of their online purchases by 2030, roughly £29 billion a year. Seven per cent is not a revolution, but it is larger than most retailers' entire marketplace channel, and it concentrates in the 18 to 34 group where 45% say they would delegate purchases.

What should a smaller retailer do about this now?

Make the catalogue readable by machines before worrying about strategy. That means structured product data with live price and stock, each product mapped to the questions and problems it answers, and a way for agents to discover it. Smaller retailers can often move faster than large ones here, because there is no legacy platform in the way.

The Takeaway

UK shoppers are not waiting to be convinced. They are the most confident in Europe about buying with AI assistance, they are already making 50 million shopping visits a month to chat platforms, and the youngest third of them are ready to hand over the purchase itself.

The constraint is on the other side of the transaction. Whether a retailer participates in that spending is decided by something far less exciting than strategy: whether a machine, asked a specific question by a real buyer, can read your catalogue well enough to name your product.