by Markus Johannes Baier · 29 July 2026
When people talk about agent readiness, they usually mean product data. Understandable, that is where the biggest lever sits. The second biggest sits one level to the side and gets overlooked almost every time: the service facts.
An agent comparing on behalf of a customer ultimately answers not just "which product" but "from whom". For that decision, the questions around the purchase matter: what does shipping cost, how long does it take, how long is the return window, how does the return work, which payment methods exist. Two shops with an identical product and price differ for an agent precisely at these points.
The catch: in most shops these answers only exist as prose, spread across FAQ pages, shipping info and terms, sometimes as a PDF. A human finds them with some patience. An agent reading the raw HTML often finds fragments, and where it has to guess, it recommends the shop where it does not have to.
The list is manageable. Shipping costs including the free-shipping threshold, delivery times per region, return window and return route, who pays return shipping, accepted payment methods and a reachable contact. All information every shop already has. The task is only to publish it in one clear place in machine-readable form: as structured data on the service pages, and consistently with that in the product markup, where shipping and return details can sit directly on the offer.
In our checks we see missing service facts less often than contradictory ones. The banner promises free shipping above 50 euros, the markup says a flat 4.90. The FAQ says 30 days for returns, the terms say 14. For humans these are blemishes. For an agent assessing trust, a contradiction is worse than a gap, because it cannot tell which statement is true.
That is why this step always includes reconciliation: one source per fact, and every surface (banner, FAQ, terms, markup) hangs off that source. It is the same logic a design system applies to components, applied to content.
In the order we recommend, service facts are step three: after accessibility and product data, before llms.txt and actionability. The effort is the smallest of the five steps because few pages are affected, usually one afternoon for the structure and a second for the contradiction check.
Whether your service facts are readable for agents is something the free AI Agent Readiness Check shows in two minutes, together with the other dimensions.