Why a company wiki is now marketing infrastructure

The pages that AI assistants quote when someone asks "what is X" are rarely blog posts. They are definitions: one term, one clear answer, stable over time, on a site whose other pages agree with it. That is what a wiki is, and it is why we built a public one before we wrote this first post.

What changed

Search used to reward the page that best matched a query and had the most links. That still matters for SEO, but a growing share of questions never reach a results page. They are answered inline by an assistant that read several sources and picked what to say. Two newer disciplines describe how to be picked:

• Answer engine optimization is about being the direct answer: a clear statement near the top of a page, with structured data that repeats it.
• Generative engine optimization is about being cited when a model synthesizes from many sources: consistent facts, attributable claims, and machine-readable copies of the page.

Both reward the same thing. Say one thing per page, say it plainly, and say it the same way everywhere.

Why a blog alone does not do it

A blog post is dated, opinionated, and long. Those are strengths for a reader and weaknesses for a machine that wants a definition. Posts also drift: the 2024 post says one thing, the 2026 post says another, and the assistant does not know which to trust.

A wiki fixes the drift. Each term has one page, the page has one description, and every post links to it instead of re-explaining it. When the definition changes, it changes in one place and every post that references it stays correct. On this site, every term in this post is a link into the HookVerb wiki for exactly that reason.

What the wiki needs to do to be quoted

The wiki is only useful to an answer engine if the engine can read it the way it prefers. Ours does four things a normal CMS page does not:

1. It serves a markdown copy of every article on request (Accept: text/markdown), so an agent gets the text without the layout.
2. It carries structured data per article: Article or DefinedTerm, with the description and the lead paragraph stated in the JSON.
3. It states its position on AI use in a header, so an engine knows it may quote and index the page but may not train on it.
4. It exposes the same content over a Model Context Protocol server, so an assistant with a connector can search it directly rather than scraping.

None of that is exotic. It is a week of work for a developer who knows the stack, and it is the difference between being a source and being a search result.

What to do this quarter

• List the fifteen terms your prospects ask about. Write one page each. A definition, how it differs from its neighbours, and what your company does about it.
• Give each page a one-sentence description a machine could quote out of context.
• Link every future post to those pages instead of redefining the terms.
• Serve a plain-text or markdown version and structured data. If your platform cannot, that is a platform problem worth raising.

FAQ

Does a public wiki give away the consulting?

No. The wiki holds definitions and how things work in general. The engagement is applying them to one company's audience, budget, and vendors, which no page can do.

Should the wiki and the blog be separate sites?

Same site, separate paths. Search engines treat the domain as one entity, and a definition on the same domain as the opinion is a stronger signal than one on a different one. Here the wiki is at /wiki and the Learning Center is at /learn.

How often should wiki pages change?

When the fact changes. A wiki page that is edited weekly looks unstable; one that is untouched for a year but still correct looks authoritative. Keep the dated commentary in posts.