Agentic commerce, explained

Strategy

• What agentic commerce means for marketplaces

  A marketplace is a comparison engine with normalised product data and identifiers, which is exactly the shape an agent wants, so marketplace listings often compete well by default. For a seller that means marketplace data quality is doing work your own site may not be, and for an independent merchant it means the competitor is a normalised record rather than a brand.

• Returns and agentic purchases

  Returns terms act as a filter during an agent’s comparison, which makes an unpublished policy an exclusion rather than a detail. After the purchase, the shopper receiving an item they did not personally select may return it for reasons your existing process does not anticipate, and the practical protection is better published specification rather than a tighter policy.

• Agentic commerce and subscription products

  An agent asked to buy something once will not commit somebody to a recurring charge, so a store that offers only a subscription may be excluded from purchases it would otherwise win. Where subscriptions are offered alongside a single purchase, the terms have to be published as facts: frequency, price per delivery, minimum term and how cancellation works.

• Agentic commerce for B2B stores

  A B2B store with gated pricing and approved accounts is rarely directly purchasable by an agent, and it is still evaluated by one. Buyers delegate research long before they delegate purchasing, so the practical requirement is publishing enough for a business buyer to shortlist you: what you sell, to whom, in what quantities, on what terms.

• How agents handle out of stock products

  An agent that is told a product is available and then cannot buy it does not try a different variant, it moves to another store. That makes accuracy in the negative more valuable than accuracy in the positive, and it is the case almost nobody tests, because the person checking is looking at a product that happens to be in stock.

• How agents compare shipping costs

  An agent comparing stores works on delivered total and delivery expectation rather than item price, because that is what the shopper asked about. A store whose shipping cost appears only at checkout cannot supply either, and it is dropped from budget or date constrained comparisons rather than placed lower in them.

• How agents use reviews and ratings

  Reviews and ratings influence an agent’s decision where they are readable as data and not at all where they are rendered by a widget after the page loads. Since review widgets are among the most common client rendered components on a storefront, many stores have accumulated strong social proof that does not count toward them in any comparison.

• Price transparency and AI agents

  Deferring costs until checkout is a long standing tactic that works on human shoppers and inverts against agents. An agent computing a delivered total treats an unknown as risk rather than as potential upside, so a store with genuinely cheap delivery and no published rule loses to a competitor with worse delivery and a stated number.

• Do upsells survive agentic checkout?

  An agent purchasing on somebody’s behalf buys what was asked for, so an upsell placed in its path is at best ignored and at worst ends the purchase. That does not mean losing the revenue: the same value can be moved to bundling, which enters the comparison as a better offer, and to post purchase, which reaches the human who actually receives the order.

• What agentic commerce does to your brand

  When a purchase is mediated by an agent, the parts of a brand that operate through feeling do not reach the decision, and the parts that resolve to checkable facts do. Warranty terms, returns generosity, provenance and specification survive the trip; photography, tone and art direction do not. That is a change in which brand investments pay in this channel, not evidence that brand stops mattering.

• What agentic commerce does to paid search

  When a shopper delegates a purchase to an assistant, the comparison can happen without a search result page ever being rendered, which is a context your bidding cannot enter. The practical implication is not to cut budget on a forecast, it is that the feed quality you already maintain for shopping ads is doing double duty, and that measuring the shift is more useful than predicting it.

• What agentic commerce does to conversion rate

  Agent traffic distorts conversion rate in both directions: requests that never run your analytics are invisible, and referred sessions arrive with intent far higher than your baseline. Reporting them together produces a number that describes no real population, so the practical response is to segment assistant referred sessions and judge them against themselves rather than against site average.

• Should you block or allow AI shopping agents?

  Blocking AI shopping agents is a legitimate commercial decision and so is allowing them. The indefensible position is the common one: a rule written for training crawlers that also removes you from a sales channel, discovered years later. The arguments on both sides are real, and the thing that matters is that the answer is chosen rather than inherited from a default.

• AI agent traffic vs bot traffic

  Non-human traffic is not one category. Search crawlers, training collectors, shopping agents and abusive scrapers have different purposes and different consequences, and the tooling most stores use groups them together. That grouping is the mechanism by which a store intending to protect its content removes itself from a sales channel.

• Agentic commerce vs traditional ecommerce

  Traditional ecommerce optimises for a human eye arriving on a page: layout, imagery, persuasion and the sequencing of a funnel. Agentic commerce optimises for a program reading data, which cannot be persuaded, does not browse, and resolves anything unstated against you. A store can be excellent at one and fail entirely at the other, and the two do not trade off as much as they appear to.

• SEO vs GEO vs AEO for ecommerce

  SEO optimises for a page that satisfies a query, GEO and AEO optimise for content that an assistant can quote and attribute, and neither is quite the job a store has. A merchant is trying to be a comparable offer, which is decided by structured product facts rather than by pages or by quotable prose, and that distinction explains why a well ranked store can lose every agent comparison it enters.