2027 will change how products are discovered and sold

Fashion retailers have long decided how customers have shopped. AI is turning that relationship on its head, explains Justin Thomas, VP Sales – EMEA North, Akeneo.

Fashion retailers built the categories, wrote the product descriptions, designed the product detail pages (PDP), set the navigation and decided which attributes mattered. Consumers learned to navigate that structure, translating what they wanted into keywords, filters and categories that a retailer’s website could understand.

Now, instead of shoppers adapting their intentions to the structure of a retailer’s ecommerce site, commerce is adapting itself to the intentions of the shopper. In 2027, we could see more shopping experiences shifting around those needs, rather than the traditional fixed sequence of static commerce, product pages, and search results.

This does not mean the ecommerce website disappears or consumers hand every purchasing decision to an autonomous AI agent. What changes is the assumption that every customer should be presented with essentially the same journey.

Imagine someone asking, “what should I pack for London in November? I’ll be walking a lot, it will probably rain and I need clothes smart enough to wear to dinner.” That is not a conventional search query; it combines weather, location, activity, occasion and personal preference. Today, the shopper does much of the work of translating those into separate searches and checking which products meet their requirements.

AI will help translate their requirements into a relevant output, using the product data from product pages and other accessible data, to produce on demand shopping experiences.  That could lead to a profound change to the role of the traditional PDP.  Images, description, specifications, reviews and an “add to basket” button still remain useful, but there is no reason every shopper needs those elements presented in the same way.

One shopper may find sustainability credentials a valuable piece of information. Another may need a comparison of technical specifications. Someone else may care about compatibility with a product they already own. AI can assemble the information and comparisons required for that particular purchasing decision.

As product experiences become more flexible – largely moulding to consumers requirements – the information behind these experiences become more important than ever. AI cannot reliably match a product or service to a consumer’s needs if the relevant data composing the agent’s decisions are unstructured, incorrect, or missing.

Much of the discussion around Agentic commerce has concentrated on consumers discovering products through external AI services, but this can obscure a bigger change: retailers’ own search engines will become conversational and contextual.

Alongside Keyword searches, Customers will increasingly describe problems, occasions and intentions. Retailers that haven’t invested in the quality, completeness and structure of their underlying product information could encounter an extraordinary problem: their products may become difficult for AI to discover even on their own websites. They require this product information to connect what an item is with the situation in which it is useful.

The competitive question will therefore shift from, “Where do I rank for this keyword?” towards, “Does an AI system understand what my product is, what makes it different and when it is relevant?”

That requires richer attributes, accurate specifications, context and relationships between products. Traditional taxonomies designed to help humans navigate product categories will still have a role, but product information increasingly needs to support machines interpreting relationships and context as well.

AI also creates a trust problem. The internet is filling rapidly with AI-generated content. Making it far easier to fabricate reviews, comparison sites and product recommendations, creating an incentive to manufacture content designed to influence AI recommendation engines, much as earlier generations attempted to manipulate search rankings.

As synthetic content multiplies, it becomes harder to distinguish independent information from material created to manipulate an algorithm. Consequently, brands and retailers can strengthen their information foundations by providing accurate, comprehensive first-party product information. Rather than assuming third-party content will fill the gaps.

Accuracy matters because trust can operate at machine scale too. Supplying misleading information no longer risks disappointing only the customer who buys the product. In an AI-mediated ecosystem, unreliable information could influence whether automated systems trust or recommend the product at all.

As AI progressively plays a larger role in the shopping experience, product and price will come together because AI assistants can evaluate both simultaneously. If two apparently similar products cost £80 and £120, an AI agent needs reliable data explaining why the expensive premium price is justified. Better materials, provenance, durability, performance, sustainability or service may provide that justification, but only if those differences are documented in the product information available to the agent.

Pricing therefore becomes part of the product data ecosystem rather than a separate calculation applied at the end. But this doesn’t have to mean personalised prices based on everything a retailer knows about an individual. In fact, AI could enable the opposite. If a retailer understands someone’s expressed intent, it has less need to infer it from an extensive customer profile. The combination of intent, rich product information and intelligent pricing can match the right product and proposition to the customer’s requirement.

In 2027, shoppers may reach the same product through different journeys, with search, product experiences, recommendations and pricing adapting to their needs. Retailers cannot control every route shoppers take, but they can ensure every experience is supported by the same requirement: product information that is accurate, structured, comprehensive and reliable enough for machines to understand.