Your customers ask ChatGPT what to buy. We ask the same questions — 48 of them, on real AI screens — and show you exactly where your brand shows up, where it goes missing, and what AI gets wrong about you.
One-time audit, delivered within 3 business days. Every score is backed by real, saved AI answers — quoted in your report.
From our pilot audits (ChatGPT, core question set, measured October 2026). Mention rate = how often the brand appears when customers ask category questions without naming any brand.
Mentioned in 76.7% of unbranded answers. A well-known brand that AI mostly gets right — accuracy 20/20, no invented facts found.
Mentioned in 26.7% of unbranded answers — and in 0 of 9 problem-based questions, exactly where purchase intent is highest. Full sample report →
Mentioned in 10.0% of unbranded answers — including 0 mentions in questions about its own local area. Beloved locally, nearly invisible to AI.
48 questions in 6 families: direct brand questions, category recommendations, comparisons, problem-based questions ("my plants keep dying"), buying & trust questions, and local/niche questions. Your brand name never appears in the unbranded ones.
We run the questions on the actual consumer interfaces of major AI assistants — the same screens your customers see — and save every answer. Each question is asked 3 times, because AI answers vary.
Mention rate (30) · Prominence (20) · Factual accuracy (20) · Sentiment & framing (15) · Citations & sources (15). Accuracy is checked line by line against the facts you confirm with us first.
Not just a score: a prioritized playbook — fix wrong facts first, then win the citations AI trusts, then build compounding presence. Every claim in the report links back to a saved answer.
The five grades: 85+ AI Favorite · 70–84 Visible Contender · 50–69 Half-Seen · 30–49 Mostly Invisible · 0–29 AI Ghost
Owners of our AI Visibility Audit Kit on Gumroad audit at the kit-owner price of $59. Mention your Gumroad receipt email when you order and we'll apply it.
Monitoring subscriptions (monthly re-measurement from $29/mo) open after launch — audit customers get first access.
You can — once. AI answers vary from session to session, so a single answer tells you very little. We ask each question 3 times, across 48 structured questions, score the answers against your confirmed facts, and show you the pattern — including the questions where you never appear at all.
The audit is measured on the consumer web interfaces people actually use, starting with ChatGPT. Coverage expands as we add assistants; your report always states exactly which assistant and conditions produced each score.
No. Every score is computed from saved, real AI answers, and your report quotes them. Where the sample is small, we show the statistical range instead of a single pretending-to-be-precise number.
About 10 minutes: your brand name, website, main market, and your key facts (founding year, HQ, flagship products, price range). We verify those facts with you before scoring accuracy — AI is graded against the truth, not against our guesses.
That's the most valuable finding in the report. Wrong facts are flagged with the exact quoted answer, and the Fix Pack drafts the corrections and the source pages AI pulls from.
If we can't complete your audit, we refund you in full. See the refund policy for details.