What's in an AI visibility report?

Q: What is share of voice in an AI visibility report?

Share of voice is how much of the answer space your brand owns versus the competitors you track, not just whether you appear. It measures your dominance across a prompt list, so a brand mentioned in every answer alongside five rivals has lower share of voice than one that owns the answer alone.

Q: Why do citations matter in an AI visibility report?

Citations show which pages an engine used to justify its answer and whether any are yours. They are the most actionable part of the report: if Perplexity keeps citing a third-party listicle you are absent from, the fix is getting onto that listicle, not editing your homepage.

Q: How many AI engines should a report cover?

All five that matter: ChatGPT, Perplexity, Gemini, Claude, and DeepSeek. Each answers the same prompt differently, so single-engine reports miss where you win and lose. Perplexity leans on fresh citations while ChatGPT often answers from trained knowledge with fewer links.

Q: Why does a report need trends over time?

A single snapshot cannot separate a real change from normal variance. Week-over-week trend lines on every metric let you prove a content or citation fix actually moved your visibility instead of guessing.

Question

What's in an AI visibility report?

Maciej Grabek · Accepted Answer

An AI-visibility report shows how often ChatGPT, Perplexity, Gemini, Claude, and DeepSeek mention your brand, where you rank against competitors, which sources they cite to justify the answer, and whether the tone is positive. The useful ones tie every number back to a tracked prompt and show the trend over time, so you know what changed and what to fix next. The core coverage metrics Four numbers sit at the top of any report. Each is measured against a fixed list of prompts, so they only mean something once that list is set. Tracked prompts - the real questions buyers type, like "best project management tool for agencies". Everything else is scored against this list, so a report with vague or invented prompts is worthless. Visibility % - the share of those prompts where your brand appears in the answer at all. Share of voice - how much of the answer space you own versus the competitors you track, not just whether you show up. Average rank - when you are mentioned, how early in the answer you land. Being named last in a list of ten is not the same as being named first. Where the answer comes from The engines do not invent answers from nothing. They pull from pages, and most will name or link the sources behind a recommendation. The citations section lists who got cited for each prompt and which source it was, and flags whether any of those sources are yours. This is the most actionable part of the report. If Perplexity keeps citing a third-party listicle to answer a prompt you care about, the fix is not your homepage. It is getting onto that listicle. Sentiment and the competitive picture Two more sections turn raw mentions into context. Sentiment - whether the mentions are positive, neutral, or negative. Being mentioned as "pricey and hard to set up" counts as visibility but is not a win. Competitor comparison - the same metrics for the rivals you track, side by side, so you can see who owns the prompts you are losing. Engine coverage and trends over time A report that only checks one engine is half a report. Each engine answers the same question differently, so coverage across ChatGPT, Perplexity, Gemini, Claude, and DeepSeek is a required column, not a nice-to-have. Perplexity leans hard on fresh citations; ChatGPT often answers from its own trained knowledge with fewer links. Then there is trends over time : week-over-week movement on every metric. A single snapshot cannot tell a real change from normal variance. The trend line is how you prove a content fix moved the needle instead of guessing. A single report row, decoded The clearest way to read a report is one prompt at a time. Take "best CRM for solar installers" tracked across all five engines - a single row of the report might read like this: Prompt: best CRM for solar installers ChatGPT: rank 4, neutral Perplexity: not mentioned Gemini: rank 2, positive Top citation: a G2 listicle you do not appear on That one row says where you are losing (Perplexity), where you are winning (Gemini), and why (a listicle you are absent from). Multiply it across your prompt list and the pattern of what to fix falls out. Recommendations: the part that matters Numbers without next steps are a dashboard, not a report. The last section should read the data and say what to do: which prompts to target, which competitor-owned sources to get onto, and which pages to build or rewrite. Good recommendations point at the specific prompt and source, not "improve your content". You can assemble all of this by hand: pick your prompts, run them through each engine, log the mentions and citations in a spreadsheet, and repeat weekly. It is tedious but doable. Tools like avisibli run the scans across all five engines on a schedule and build the report for you, so the weekly refresh is not a manual chore. avisibli is the GEO platform that publishes this answer library. Self-references are limited to topics where a tool-based answer is genuinely useful to readers.