Free business tools
Free AI Visibility Self-Audit Toolkit
Run consistent prompts. Record the evidence. Track your brand visibility.
02 / Your observed visibility
Manual audit score
Not audited
Generate a pack, then run and record the prompts.
This toolkit does not query AI engines or scan your website. You run prompts manually and record observed answers. The score reflects your sample, not a universal ranking.
Make your brand easier to understand.
Discuss your content, evidence and website structure with Dawood.
Review my AI visibility →03 / Your prompt pack
Run each discovery prompt in a separate new chat on ChatGPT, Perplexity and Gemini. Use the same search setting, location and wording on repeat audits. The brand is intentionally absent from discovery prompts to avoid prompting a mention.
04 / Record the evidence
For each answer, choose a result and paste a source link or answer excerpt. A result without evidence remains unreviewed. Check citation only if the answer cites your official domain, and accuracy only after checking material claims. Record model/version, date and search mode below.
| Engine / prompt | Observed result | Official citation | Claims accurate | Evidence / answer excerpt |
|---|
A repeatable GEO audit
How does the 0-100 score work?
Each evidence-backed answer can earn four equal points: brand mentioned, explicitly recommended, official domain cited, and material brand claims accurate. Recommendation also counts as a mention. Citation and accuracy points only count when the brand is mentioned. An absent brand earns zero. Score = earned points / (4 x reviewed answers) x 100, rounded. This is our transparent audit rubric, not an engine-provided score.
Unrun answers and answers without evidence are excluded, not treated as failures. The score remains provisional until all 15 answers are reviewed. Compare only completed audits with the same prompts and settings; repeat runs to assess variability.
How do I avoid biased checks?
Use unbranded discovery prompts first. Start a new conversation per prompt, record engine/model and search mode, and avoid introducing your brand in earlier messages. The pack includes a separate branded diagnostic prompt; it is excluded from scoring because naming the brand biases mention results. Record competitors in your evidence notes if useful.
What should I improve after the audit?
Check factual errors against your official website. Make service, location, contact and supporting evidence clear and consistent. Publish useful, attributable answers to real customer questions. Validate that important pages can be accessed and indexed. No change guarantees a mention in generated answers.
Practical guide
What this tool helps you decide
This free AI visibility toolkit helps you run a repeatable manual audit of how AI answers describe and recommend your business. It generates discovery prompts and a scoring sheet for brand mentions, recommendations, official citations and factual accuracy.
Use it for a baseline generative engine optimization (GEO) audit, a brand-content review or a repeat check after improving your website. You run the prompts yourself; the toolkit does not query engines or perform an automated visibility scan.
How to use the AI visibility self-audit toolkit
Describe the brand and market
Enter your brand, official domain, service category and target location. Generate the prompt pack and a new scoring sheet. Export existing work before generating again because a new pack clears the current sheet.
Run discovery prompts in fresh conversations
Run each of the five unbranded prompts separately in ChatGPT, Perplexity and Gemini. Keep wording and settings consistent. Do not introduce your brand first, because that would bias the mention check.
Record the result and supporting evidence
For each of the 15 answers, record whether the brand is absent, mentioned or recommended. Paste an answer excerpt or evidence link. Check official citation and accuracy only after inspecting the answer and its material claims.
Complete the sample and export it
Record date, model/version, search mode and region. Finish all 15 evidence-backed answers before comparing scores. Download the prompt pack and CSV sheet, then repeat the same protocol on a later audit.
Worked example
Imagine a completed audit of five prompts across three engines. Six answers mention and recommend the brand, cite its official domain and contain accurate material claims. The other nine answers do not mention the brand.
- Evidence-backed answers
- 15 / 15
- Points earned
- 24 / 60
- Completed sample score
- 40 / 100
Illustrative scenario, not a forecast or service quote. Change the inputs in the calculator to match your circumstances.
How to interpret your results
The score measures the answers you observed under a specific protocol. A high provisional score from one answer is not the same as a completed 15-answer audit. Use the evidence to distinguish missing mentions, incorrect facts and weak source coverage rather than treating the number as an AI ranking.
Common input mistakes to avoid
- Do not score the branded diagnostic prompt as an unbiased discovery result.
- Do not mark a citation as official merely because it mentions the brand; check the cited domain.
- Do not compare different prompts, engine settings or incomplete samples as if they were equivalent.
Questions before you use the results
Is this a free AI visibility checker?
It is a free manual self-audit toolkit. It generates prompts and calculates a score from your observations; it does not automatically check live mentions in ChatGPT, Perplexity or Gemini.
Why are some answers missing from my score?
Unrun answers and answers without recorded evidence are excluded. Completion is shown separately, and the score is provisional until all 15 discovery answers have evidence.
Can I use this to track improvements over time?
Yes. Save the prompt pack, completed sheet and run context, then repeat the same prompts and settings. Generated answers vary, so repeat observations and inspect the underlying evidence alongside the score.