The AI Visibility Index
Q: What is the AI Visibility Index?
A: An open benchmark measuring how often the major AI engines name and cite brands within a category, scored on mention rate, citation rate, and average position across six engines under neutral conditions.
An open benchmark of how often AI engines name and cite brands, by category. Published prompt sets, published methodology, six engines, neutral conditions — so the numbers mean something to someone who did not run them.
- Six engines, one fixed prompt set per category
- Methodology, prompts, and weightings published with every edition
- Raw evidence available to any brand that asks for its own
Open methodology · CC BY 4.0 · Sophyx excluded from its own rankings
Edition one is being measured now
We are not going to publish numbers before we have run them. The methodology below is final and the first categories are in measurement; results publish here when the run completes, along with the full prompt set so anyone can reproduce it.
If you want your category covered in edition one, or want your own brand scored against the same method today, the free analysis uses the same measurement:
Methodology
Published in full, because a benchmark nobody can reproduce is an advertisement.
Prompt set
Each category gets a fixed set of buyer-intent prompts written the way a real purchaser would phrase them — no brand names in the prompt, because including one guarantees the mention we are trying to measure. The set is published with each edition so anyone can reproduce it.
Engines
Every prompt runs against all six engines: ChatGPT, Google Gemini, Google AI Overviews, Claude, Perplexity, Microsoft Copilot. Running a subset and calling it an index would bias the result toward whichever engine a brand happens to have optimised for.
Neutral conditions
No personalisation, no logged-in accounts, no location signals beyond a stated default. Personalised results are real for the individual user but useless as a comparable benchmark.
The score
A 0–100 composite of three measured quantities: mention rate (how often the brand is named), citation rate (how often the brand's own domain is cited in answers that name it), and average position among brands named in the same answer. Weightings are published with each edition rather than kept proprietary.
What we publish
The scores, the prompt set, the engine list, the run dates, and the weightings. Raw per-answer captures are retained so any brand can request the evidence behind its own score.
Conflict of interest
Sophyx sells AI visibility software and builds this Index. Sophyx does not appear in its own rankings. If a scored brand is a Sophyx customer, that is disclosed in the edition footnote rather than in the score.
Score your own brand on the same method
The Index and the free analysis use the same measurement, so your score is comparable to the published editions rather than a separate scale. Check a single engine, or all of them at once:
Free Perplexity checker
Same method, one engine, no signup
Free Gemini & AI Overviews checker
Same method, one engine, no signup
Free Claude checker
Same method, one engine, no signup
Free Microsoft Copilot checker
Same method, one engine, no signup
Engines covered by the Index: ChatGPT, Google Gemini, Google AI Overviews, Claude, Perplexity, Microsoft Copilot.
AI Visibility Index FAQ
- An open benchmark measuring how often the major AI engines name and cite brands within a category. Each edition runs a fixed, published set of buyer-intent prompts across ChatGPT, Gemini, Google AI Overviews, Claude, Perplexity, and Microsoft Copilot under neutral conditions, and scores each brand on mention rate, citation rate, and average position.
- Scores are only meaningful within a category, because category norms differ enormously — a 40 in a crowded consumer vertical can represent a stronger position than a 70 in a niche with three players. Compare against the other brands in your own category and against your own previous edition, not against an absolute threshold.
- A grader scores one brand you submit. An index measures a whole category on a fixed prompt set and publishes the comparison, so the numbers mean something relative to each other. Both are useful and they answer different questions: a grader tells you your score, an index tells you your rank.
- Yes. We retain the per-answer captures behind every score and will share the evidence for your own brand on request. An index nobody can audit is a marketing asset, not a benchmark.
- No. Sophyx builds and publishes the Index and sells AI visibility software, which is a conflict of interest we manage by excluding ourselves from every category we score. Where a ranked brand is a Sophyx customer, that is disclosed in the edition footnote and does not affect the score.
- Quarterly. AI answers move weekly, so a single snapshot is close to worthless — the value is entirely in the trend line across editions. If we cannot sustain the cadence we will say so rather than quietly let it go stale.
Where does your brand sit?
Free, no signup. Mention rate, citation rate, and position across six engines — the same three numbers the Index scores.
AI share of voice · AI citation tracker · Why AI skips your brand · Compare the tools