Loading...
The GEO playbook Moonrank's GEO Agent uses to measure how often ChatGPT, Perplexity, and Gemini mention your brand, and turn the trend into a content backlog.
Download the methodology, see how the agent builds your prompt set, or run it live inside the GEO Agent.
--- name: ai-visibility-tracking description: Build the prompt set that measures how often AI assistants mention your brand, and turn the trend into a content backlog. owner: Moonrank GEO Agent --- # AI Visibility Tracking Rankings tell you where a page sits. This tells you whether an assistant says your name. Build the prompt set, measure it on a schedule, act on the drops. ## When to use - You need a number for "are we visible in AI search?" that you can report monthly. - You ship GEO content but can't prove it moved anything. - A competitor is suddenly cited everywhere and you want to know where. ## Inputs - **Required:** your brand name, your domain, and your category. - **Optional:** competitor brands to benchmark against, and target buyer personas. ## Method 1. **Prompt set:** write 30–50 prompts a real buyer would type, spanning discovery, comparison, and objection questions. 2. **Segment:** tag each prompt by funnel stage and persona so the trend is readable. 3. **Run:** query each engine on a fixed schedule, same wording, same locale. 4. **Score:** for every response record mentioned yes/no, position in the list, sentiment, and the cited source. 5. **Attribute:** map each mention back to the page or third-party source that earned it. 6. **Act:** prompts where you're absent become the content backlog; prompts where you slipped become the refresh list. ## Output format A visibility report: - **Mention rate**: % of prompts where the brand appears, per engine - **Share of voice**: your mentions vs each competitor - **Sentiment**: positive | neutral | negative - **Source**: the URL the engine cited, if any ## Prioritization rules - Never change prompt wording mid-series; you lose the trend. - Weight bottom-funnel prompts higher — "best X for Y" converts, "what is X" does not. - Chase the cited source, not just the mention: owning the source is what compounds. - One engine moving alone is noise; three moving together is a signal.
The AI Visibility Tracking skill is how Moonrank's GEO Agent turns a vague question — are we visible in AI search? — into a number you can report and a backlog you can work.
It's written as a SKILL.md, a plain-language instruction format that people and AI agents can both follow. Read it to build your own prompt set and scoring rules, or let the GEO Agent run the measurement on a schedule and tell you what moved.
Downloading the playbook is free with a Moonrank account. Running the prompt set continuously across engines, and attributing every mention back to a source page, is what the GEO Agent does for you.
“We shipped twelve GEO articles last quarter. Did any of it work?”
Mention rate went from 18% to 31% across 44 prompts, driven almost entirely by comparison queries — your three new versus pages are now the cited source in nine of them. Discovery prompts didn't move: you're still absent from every "best tools for" list. Next: two category roundups, then re-measure in three weeks.
Your brand name, domain, and the competitors you want to be measured against.
30–50 prompts a real buyer would type, tagged by funnel stage and persona.
Same wording, same locale, every engine — so the trend is real and not noise.
Prompts where you're absent become the backlog; prompts where you slipped become the refresh list.
Discovery, comparison, and objection questions — not just brand-name lookups.
The headline number, broken out by ChatGPT, Perplexity, Gemini, and AI Overviews.
Your mentions measured against the competitors you actually lose deals to.
Being mentioned badly is not the same as being mentioned well.
Which page or third-party source the engine actually cited.
Fixed wording and cadence, so month-over-month movement means something.