AIgentSphere

Glossary › Fact block

Fact block — the answers, in the bytes of the page an agent reads

A fact block is a plain-text panel of buyer-question answers placed in the homepage itself, so an agent that fetches one page gets the facts on that fetch.

This is our own term for the measured artifact, and it is the only surface in the package with a controlled result behind it. The construction that won: wrap the paragraph the homepage already shows under its headline in a collapsed panel, and put the answers inside it. A visitor sees no change. An agent reading the bytes gets a labelled list of answers.

Three properties are load-bearing, all of them learned by failing first. The facts are plain HTML text, never JSON in a script tag — the common text extractor strips scripts before a model ever reads, so data in a script tag is measured dead (Note 39). It sits near the top, because on a long page a bottom-placed block stops arriving at all (Notes 36, 37). And it answers the questions buyers ask rather than listing specifications (Note 38).

What we measured

Same site, same six models, same day: 78% of fact checks recovered with the block against 30% in the control, mean crawl depth 1.1 pages, and no rise in stated confidence — 0.86 to 0.85 (Notes 30, 41). Facts deliberately left out of the block stayed unrecovered, with no halo effect.

Source: MARKETING_NOTES_AGENT_VISIBILITY.md. Every note number is a dated measurement with its method and its cost.

disambiguation

What Fact block is not

free audit

See what AI sees on your site

Vocabulary is cheap. The measurement is the part that tells you whether any of it applies to your site.

One model fetches your homepage and answers buyer questions from it. You get the recovery score and the specific answers it could not find. That is the whole free tier: it shows you the problem, and the fix package is not included.