Loading...
The playbook Moonrank's GEO Agent uses to turn AI-visibility checks into a report someone can act on — with the prompts where the brand is invisible listed first.
Download the methodology, see the prominence scale, or run it live inside the GEO Agent.
The AI Visibility Report skill is how Moonrank's GEO Agent reads the results of its own visibility checks. It keeps two numbers strictly apart — how often the brand is named in an answer, and how often its domain is cited as a source — because a brand can be mentioned constantly and cited never, and that is a different problem from the reverse.
It breaks both down by engine and by prompt, flags every prompt where the brand is invisible across all engines, maps which domains the engines cite instead, and computes share of voice against competitors. The actions at the end are tied to those findings: a prompt to target, a source to earn a placement on, a competitor citation to match.
It's written as a SKILL.md, a plain-language instruction format that people and AI agents can both follow. Read it to understand what the AI Visibility score is made of, hand it to a client as the method behind their number, or let the GEO Agent produce the report from your workspace.
“Our AI Visibility score is 14%. What does that actually mean and what do I do?”
Mention share is 22%, citation share is 8% — you get named more than you get cited, so the engines know you exist but don't trust you as a source. Six of your 18 prompts are invisible on every engine, all in the comparison cluster. Gemini cites you on three prompts; ChatGPT on none. G2 and one competitor's blog are the two most-cited domains for your topics. The first action is a comparison page built to be cited, not ranked.
Every tracked prompt's most recent result per engine, with failed checks excluded rather than zeroed.
Two shares, never merged except inside the clearly labelled AI Visibility Score.
Where the brand is strong, where it is weak, and every prompt where it is invisible.
Who the engines cite instead, competitor share of voice, and actions tied to each gap.
Kept separate throughout, because they fail differently and are fixed differently.
Named first, shortlist, passing mention, absent — the only judgement call in the report.
Every prompt with zero mentions and zero citations across every engine.
Mention rate, citation rate and modal prominence for each engine tracked.
The domains and pages the engines actually cite, with the brand's own rows marked.
Earning citations and source placements — tied to findings, not recycled SEO advice.