Agentic AI is moving discovery closer to the transaction 

For years, ecommerce has treated discovery and conversion as two separate jobs. First, a shopper searches, then they browse, compare, read reviews, abandon a few tabs and eventually decide whether to buy. That journey is beginning to compress.

AI assistants are changing the role of discovery from a list of links into an active conversation. Instead of asking a shopper to navigate a catalogue, they can interpret what the shopper is trying to achieve, narrow down the options and guide them towards a decision. 

The new starting point is intent

Agentic AI is moving discovery closer to the transaction. It is not simply helping people find products. It is helping them resolve the decisions that sit between a need and a purchase.

McKinsey’s thinking on agentic commerce describes a future in which AI agents can operate within rules set by the customer, including budget, delivery windows, preferred retailers and acceptable substitutes. Under moderate scenarios, McKinsey estimates that AI agents could mediate between $3 trillion and $5 trillion of global consumer commerce by 2030.

The important word here is “mediate”.

Agents may not complete every transaction autonomously. But they can increasingly influence which products make it into the consideration set, which means retailers need to think beyond traditional rankings.

The question is no longer only: “How do we rank first?” it’s also: “Can an AI agent understand why our product is the right choice?” 

Product data is becoming part of the customer experience

This shift creates a practical challenge for retailers. An AI assistant cannot recommend what it cannot understand. Product names, descriptions and images are no longer enough on their own and retailers need structured, accurate information about ingredients, materials, sizing, use cases, delivery windows, returns, availability, pricing and suitable alternatives.

They also need a clear proposition. When a shopper asks an AI assistant to find the best product for a specific goal, the brand with the loudest campaign may not win. The product that is easiest to understand, compare and trust may have the advantage.

That puts pressure on every part of the commerce operation. Product information needs to be machine-readable, stock needs to be reliable, fulfilment promises need to be accurate and returns policies need to be transparent. In other words, agentic commerce is not just a marketing opportunity, it tests hte entire operation.

If a retailer’s catalogue, policies and value proposition are not clear enough for agents to interpret, the retailer may be invisible at the precise moment a shopper is ready to decide.  

Retail media needs to become more useful

There is a further implication for retail media, if an AI assistant becomes the place where a shopper compares products, sponsored recommendations cannot simply interrupt the conversation. They need to contribute to it.

That is the thinking behind THG Commerce’s partnership with Criteo and Google Cloud.

THG Commerce has integrated Criteo Sponsored Products into its AI Shopping Assistant, built with Google Cloud’s Gemini Enterprise. Sponsored products appear alongside organic recommendations within the conversation, using the shopper’s own language and requirements to determine relevance. They are clearly labelled and only appear when they meet the relevant criteria.  

he early results are encouraging, sponsored products surfaced through the assistant achieved a click-through rate more than five times the average, with conversion rates more than twice the average. That performance makes sense... a shopper who has explained what they need is not browsing passively, they’re already signalling intent.

The opportunity for retail media is to meet that intent with something genuinely helpful: a product that fits the brief, appears at the right moment and makes the next decision easier. 

The future belongs to retailers that are easy to understand

Agentic commerce won’t reward retailers simply for adding a chatbot to their website, it’ll reward retailers that have done the harder work underneath: improving product data, connecting stock and fulfilment information, clarifying policies and making their proposition easy to interpret.  

Discovery is moving closer to the transaction because AI is removing the distance between “What am I looking for?” and “Is this the right product for me?” and for retailers, the task is clear. Make the product easy to find, make the value easy to understand, and when the moment of decision arrives, make the recommendation worth trusting.