Buyer’s checklist
What an AI visibility audit should include
“AI visibility audit” is a new label on products that vary enormously — from a quick automated scan to a multi-week engagement. Before you buy any of them (including ours), this is the checklist we’d want in your hand. Each item says what to look for, then how CiteMe’s $495 Pro run handles it.
01Tests non-branded buyer questions, not just your brand name. Brand-name prompts only test whether AI knows you exist; buyer questions test whether it would recommend you.
CiteMe Pro run: The Pro run tests 105 buyer questions across four types: 30 best-fit, 30 alternative-provider, 30 use-case, and 15 community questions. Most are written without your company name; the alternative-provider questions name your company explicitly, to show how engines position you when a buyer is already comparing.
02Shows the exact questions, engines, dates, and sample sizes. A number without its inputs can’t be checked or re-run.
CiteMe Pro run: Every result shows the exact question, the engines, the run date, and the sample size. How CiteMe queries engines, exactly as documented on the methodology page: ChatGPT — the gpt-4o model with OpenAI’s web search tool enabled, searching the live web before answering. Gemini — the gemini-3.8-flash model with Google Search grounding enabled; the model decides question by question whether to search. Perplexity — the sonar-pro model, which is search-native. Answers reflect exactly those configurations, and API results can differ from what the consumer apps show.
03Shows full answers, not only a score. The context around a mention — or an absence — is where the useful information lives.
CiteMe Pro run: You get the full answer text for every question and engine, not just whether you were named.
04Captures the sources shown with each answer. Visible links, directories, and articles are your clues about where answers come from — a clue, not proof of an engine’s reasoning.
CiteMe Pro run: Sources are recorded as each engine reports them, question by question, so you can see which directories, review sites, and articles keep appearing.
05Takes repeated samples instead of one answer. Answers vary run to run, and a single sample overstates whatever it happens to show.
CiteMe Pro run: Each question is asked three times per engine, so stable patterns separate from one-off noise.
06Covers work outside your website. Engines lean on directories, communities, and coverage well beyond your domain.
CiteMe Pro run: The Pro run includes source, directory, and community mapping, plus a source outreach playbook for the off-site work.
07Delivers drafts you can ship, not just findings. Output your team can act on without buying a second engagement.
CiteMe Pro run: Drafts your team can ship: FAQs, direct-answer blocks, llms.txt and JSON-LD, and a service brief — plus CSV exports, up to 5 team seats, and priority support.
08Can be re-run against the same questions later, so you can measure whether your fixes moved anything.
CiteMe Pro run: Every question is shown verbatim in the report, so the same set can be tested again — by you or by us — after changes ship.
09Makes no guarantees. Answers change with the engine, model, wording, location, and date, and no one controls what a model says.
CiteMe Pro run: CiteMe shows what happened in the runs and what’s worth fixing. It does not promise rankings, mentions, or recommendations.
Red flags when buying an audit
- Promises of specific positions or outcomes — no one controls what a model says.
- “Number one in ChatGPT” claims — the marketing version of the same promise.
- A score with no published method — if you can’t see the questions, engines, and dates behind a number, you can’t check it.
- Repackaged SEO — an audit that never shows an actual AI answer or source is a site audit with a new label.
See what the engines say before you buy anything
The free snapshot runs one buyer question through all three engines. The Pro run does the full checklist above.