AI Shopping Agents Are Starting to Send Real Customers to Retailers

01 Event

AI shopping agents are starting to send measurable customer traffic to major retailers. John Lewis, Britain’s largest employee-owned retailer, said searches from AI agents reached 2.5% of product searches, up from just 0.3% a year earlier, according to Reuters.

The share is still small in absolute terms, but the growth rate is striking. A move from 0.3% to 2.5% means AI-agent searches became more than eight times as common in one year.

John Lewis is responding by increasing investment in content creation. The retailer opened a studio inside its Oxford Street flagship where influencers can record content and is producing more of its own video programming. The idea is not only to reach people directly but also to create information that AI systems can discover, interpret and surface.

The trend comes while consumers remain cautious about discretionary spending. John Lewis executives described the economy as difficult, with shoppers concerned about inflation, interest rates and employment prospects.

02 What Changed?

The important change is that AI shopping is moving from a futuristic concept into measurable referral behavior. Consumers are increasingly asking AI systems to research, compare and recommend products rather than beginning every shopping journey with a traditional search engine or retailer website.

An AI agent can potentially do more than answer a question. It can compare specifications, summarize reviews, identify retailers, narrow choices and eventually complete parts of the purchase process on a user’s behalf.

That changes what retailers need to optimize. Traditional ecommerce strategy focuses heavily on search-engine rankings, paid ads, email, marketplaces and social media. AI-driven discovery adds another layer: can an AI system understand the retailer’s product information well enough to recommend it?

John Lewis’s response suggests retailers believe richer product content matters. Clear specifications, authoritative descriptions, video, creator content and structured information can all help AI systems understand products and match them to user needs.

The trend is also supported by suppliers. Reuters reported separately that Anthropic has introduced blueprints to help retailers build shopping and merchant agents using Claude, showing that AI companies are actively targeting commerce workflows.

03 Why It Matters

Product discovery is economically valuable because whoever controls discovery can influence which brands receive attention. Search engines built enormous advertising businesses around this position. Marketplaces such as Amazon also became powerful because shoppers often start directly on their platforms.

If consumers increasingly begin with AI assistants, some of that influence could shift again. Retailers may become dependent on how AI systems interpret product data, reviews, availability and brand reputation.

This creates opportunity for smaller brands as well as risk. A good AI recommendation system could surface a lesser-known product because it matches a user’s needs better. But brands with poor data, weak content or limited online evidence may become harder for agents to understand.

Retailers also face a measurement challenge. Traditional web analytics can track clicks, campaigns and referral sources. AI agents may summarize information without sending a click every time, making attribution more complicated.

The John Lewis numbers are therefore important even though 2.5% is still a minority. The trend is visible enough for a major retailer to invest in adapting its content strategy.

04 What It Means for You

For shoppers, AI agents can reduce research time. Instead of opening many tabs, a user can ask for a shortlist based on budget, size, features or other constraints.

But convenience creates a new risk: the user may rely on the agent’s summary without checking the underlying retailer information. AI systems can misunderstand product specifications, use outdated data or fail to recognize regional differences.

Consumers should therefore treat AI shopping advice as a starting point rather than an unquestioned purchase decision. Verify price, warranty, return policy, compatibility and delivery terms on the seller’s own site before buying.

For retailers, the practical task is to make product information machine-readable and trustworthy. Clear titles, accurate specifications, structured data, high-quality images and consistent availability information are becoming more important.

For publishers and affiliate sites, AI shopping agents create both competition and opportunity. Some informational searches may be answered directly by AI. But original comparisons, testing, structured buying guides and reliable niche expertise can become sources that AI systems cite or learn from.

05 Numbers + Context

John Lewis said AI-agent product searches increased from 0.3% to 2.5% in one year. That is an increase of 2.2 percentage points.

Relative to the starting point, the share increased by more than eight times. That sounds dramatic, but context matters: 97.5% of product searches were still coming through other routes.

This is a classic case where percentage growth and absolute market share tell different stories. An eightfold increase can be strategically important even while the channel remains small.

Imagine a retailer receives 10 million product-search visits or interactions a year. At 0.3%, AI agents would represent 30,000. At 2.5%, they would represent 250,000. That is an increase of 220,000 interactions even though the AI channel still accounts for only one in forty searches.

The example is illustrative and does not describe John Lewis’s actual traffic volume. It shows why a retailer might invest early rather than wait until AI shopping becomes a dominant channel.

The timing also matters because major AI companies are building commerce tools. Anthropic’s retail-agent blueprints show that infrastructure for agentic shopping is developing alongside consumer behavior.

06 Earnyx Takeaway

AI shopping agents do not need to dominate ecommerce to matter. They only need to influence enough high-intent shoppers to change how retailers compete for discovery.

The John Lewis data suggests that transition has started. The channel is still small, but it is growing fast enough to deserve real investment.

For consumers, the upside is less research friction. The downside is a new layer between you and the seller. That makes verification more important, not less.

For retailers, “SEO” is becoming broader than ranking in Google. Product information now needs to work for search engines, marketplaces, social platforms and AI agents at the same time.

The companies that benefit most may not be the ones producing the most content. They may be the ones producing the clearest, most trustworthy, best-structured information that machines and humans can both use.

Sources: Reuters, September 3, 2026; Reuters reporting on Anthropic retail-agent blueprints, September 2, 2026.