Agentic commerce changes the buying journey

Ecommerce has spent decades optimising the funnel for human behaviour. Retailers have invested in search, social, homepages, UX, personalisation, testing and conversion tactics designed around how people browse, compare and decide.

The session from nShift explored what changes when the buyer is no longer only human.

In an agentic commerce model, a customer may give an AI agent a goal: find the right product, stay within budget, deliver by a certain date and make sure the returns process is acceptable. The agent then compares options using available data before guiding or completing the purchase.

That shift could make the buying journey less like a funnel and more like a shorter decision path, where the agent arrives with much of the evaluation already complete.

Delivery becomes part of the selection criteria

When AI agents compare options, delivery may become more visible as a differentiator.

The agent is unlikely to be influenced by visual design, urgency tactics or traditional conversion mechanics in the same way a human shopper might be. Instead, it may look for facts: delivery date, delivery method, carrier choice, pickup options, returns policy, reliability and customer experience signals.

For retailers, this means delivery cannot sit below the line as an operational afterthought. It becomes part of the total offer alongside product and price.

Clear delivery promises, accurate estimated delivery dates and meaningful carrier choice may all become more important in helping agents decide which retailer is most likely to meet the customer’s goal.

Proof of experience matters

The discussion also highlighted the importance of public proof.

AI agents draw on information available across the internet, including reviews, forums and third-party sources. That means the delivery experience customers talk about publicly may influence how future agents assess a retailer.

A strong delivery experience can create positive signals. A poor one can create negative signals that are easier for agents to find and repeat.

This creates a potential flywheel. Retailers that deliver well generate better proof, which may help agents recommend them more often, creating more opportunities to deliver well again. The reverse can also happen if poor experiences become part of the public data that agents use.

Websites need to become machine-readable

Another practical theme was AI readability.

Retailers may need to understand how their sites, checkout flows and delivery information appear to AI agents and bots. If delivery options, carrier details, return policies or service promises are difficult for machines to read, agents may not be able to evaluate the full offer accurately.

This does not mean replacing human-focused ecommerce. It means ensuring that important information is structured, clear and accessible to both people and machines.

What this means for the DELIVER community

Agentic commerce is still emerging, but the implications for ecommerce and logistics are already worth considering.

For retailers, delivery information should be treated as decision-critical data, not just checkout content. For logistics and delivery partners, carrier performance and customer experience may increasingly influence acquisition as well as fulfilment. For technology providers, the opportunity is to help retailers make delivery choice, promise and proof more visible in agent-led journeys.

If AI agents become part of how customers discover and buy products, delivery will need to work for both audiences: the human receiving the parcel and the machine helping them choose where to buy it.