How much does an agent need to know before it can buy?
Not everything. An AI agent working through a booking request doesn't need a business's full history, it needs the small number of facts that resolve the specific thing the person asked for. A page with forty things to say and none of them the right one loses to a page with three.
6 min read
Matching is fit-testing
Two cottages near Buxton take dogs. One page opens with three paragraphs on the building's Georgian history before it mentions pets at all. The other says, plainly, near the top: two dogs, any size, boot room with a hose for muddy walks back from the Peaks. A family booking a walking weekend with a pair of spaniels doesn't need the Georgian history to decide. They need the second page's three facts, and they need them where an agent can find them without inferring anything.
An agent working through a booking request isn't searching for everything a business has ever said about itself. It's checking a short list of facts against a specific ask: these dates, this party, this constraint. "Known" means the facts that gate the decision are present, current, and findable, not that everything about a business has been written down somewhere.
The muddy-boots test
Take the two-dog cottage again. The decisive facts aren't romantic ones: how many dogs, any size limit, is there somewhere to deal with wet, muddy kit before it goes in the car. Get those three right and an agent can confirm the match in one pass. Leave them implied, "dog-friendly" with no detail, and the agent either has to ask a follow-up question the family didn't want to answer, or passes on the property rather than risk recommending a mismatch.
The same pattern holds for a very different kind of booking. A walking-tour operator listing a trip as "suitable for most fitness levels" tells an agent nothing about whether it's a sensible match for someone recovering from a knee injury. A specific note, up to ninety minutes, one steep section, an alternative flatter loop available, resolves the actual question in one line, where the vaguer phrase would have needed a follow-up call to settle.
Past a point, more content buries the decisive fact
The instinct runs the other way. A business assumes the fix for being overlooked is writing more, another paragraph, another gallery, another section of the site. Past a point, that's the opposite of useful: the decisive fact gets buried under prose that reads well to a person scrolling and does nothing for an agent checking a specific constraint. Volume isn't the same job as precision, and an agent rewards precision because that's what lets it commit to a recommendation rather than hedge one.
Where this shows up for a business
Selfe's structured record of a business isn't built as a content form asking it to describe itself exhaustively. It captures the specific facts that resolve real decision points, verified against the business's own systems rather than taken as marketing copy at face value, the same distinction "Being found is not the same as being bought from" makes at the level of the whole idea. A business with three checkable facts in the right place beats one with forty unchecked ones, every time an agent is deciding rather than just browsing.
Does a longer "about us" page help an agent match us better?
Not on its own. What helps is whether the specific facts a request depends on are present and checkable, however long or short the page is.
How do we know which facts actually matter?
They're the ones a real booking request would hinge on: a hard limit, a specific date-sensitive detail, anything that would otherwise need a follow-up question to confirm.
The things an agent cannot guess.
Getting found and getting matched are two different jobs. A business can nail the first and still lose the second, because an AI agent only knows what it's been told, never what it could have worked out by being there.
The first thing an agent gets wrong about your business.
An AI agent rarely gets a business wrong out of carelessness. It gets it wrong because the fact that would have changed the answer was never written down anywhere it could check.
Structured data, explained without the jargon.
A returns policy that says "just get in touch if there's a problem" and one that says "returns accepted within 30 days, unworn, tags attached" can describe the same shop. Only one of them can be checked by something that isn't a person using judgement.