GEO — showing up in AI answers

Ranking on Google gets you a link somebody might click. Being cited by an AI assistant gets you named in the answer itself, before any link is considered. The two aren’t the same, and they don’t move together — engines cite sources that explain a category well, which is often a roundup, a community thread, or a review you don’t control rather than your own homepage.

That’s the actionable part: you can’t optimise your way into an AI answer the way you’d optimise a page. You get there by being present in the sources those engines already read.

Visibility score — the percentage of AI answers that mentioned you, across every prompt and every engine. Twenty prompt-engine pairs with you named in five of them is a score of 25. An engine that failed to answer is skipped rather than counted against you, so one flaky provider can’t drag the number down.

Prompts where you were cited — how many of your questions had at least one engine cite a source inside your chosen scope. Counted per question, not per link, because “cited for 3 of 5 questions” is the fact you act on.

The two come apart, and the gap is informative: being mentioned without being cited means the engines know you but are learning about you from somebody else’s page.

Per question:

• Which engines mentioned you, and how many times each.
• Every source cited, categorised as editorial, community, reference, competitor, or your own — so you can see whether the answers are being built from journalism, forums, or your marketing site.
• Which other brands came up, and how often across questions.
• Recommendations, each ranked by likely impact and the effort involved.

Across all questions: the sources cited most often. Those are the pages worth being on.

Recommendations come in two kinds, and Featured drafts for both:

• Pitch — an engine is citing an article you’re absent from. Featured writes a pitch email to its author, referencing that specific piece, or a shorter LinkedIn message when email isn’t available.
• Post — the engines are learning from a community thread. Featured writes a comment, either from your perspective as someone who uses the product or transparently as someone on the team. It asks which before writing, because the two read very differently.

Featured needs your name and title for a draft, and will ask once.

One scan is a snapshot. The GEO / AI Visibility workflow re-runs the same brand on a schedule and delivers the report to email or Slack, so you see the score move. Unlike other workflows it deliberately doesn’t de-duplicate — re-measuring the same brand is the whole point. See workflows.