The Squoze AI Visibility Score
A transparent 0–100 measure of how often AI assistants recommend a brand. We publish the exact formula so you can audit it, reproduce it, and trust the number — not take it on faith.
Version 1.0 · Last updated 2026-07-12
What the score answers
Classic SEO metrics answer “where do I rank on a results page?” That question is disappearing. When a buyer asks ChatGPT, Claude, Gemini or Perplexity for a recommendation, there is no results page — there is a paragraph naming two or three brands. The AI Visibility Score answers the new question: when your buyers ask AI, how often does it name you?
How we collect the data
We simulate real buyers. For a given brand we generate a set of neutral, buyer-intent questions — the things a customer actually asks (“what's the best X for Y?”, “who should I use for Z near me?”) — phrased so a good answer would naturally name brands, and engineered not to mention the brand under test.
- Each question is put to multiple assistants — ChatGPT, Claude, Gemini and Perplexity.
- The models are grounded in live web search, so answers reflect the current web, not stale training memory.
- Questions are sampled repeatedly, not asked once — AI answers vary by phrasing, date and randomness, so a single ask proves nothing.
- We record, per answer, whether the brand was named, whether its domain was cited, where it appeared in any ranked list, and which competitors and sources were named.
The formula
Every answer is scored on four signals. The blended score is then scaled by provider coverage:
| Signal | Weight | What it captures |
|---|---|---|
| Mention rate | 45% | How often the brand is named at all across the tested buyer questions. |
| Citation rate | 22% | How often the answer links to or cites the brand's own domain as a source. |
| Recommendation position | 18% | When the brand appears in a ranked list, how near the top — first place earns full credit, decaying by rank. |
| Share of AI voice | 15% | The brand's mentions versus competitors' mentions for the same questions. |
The four weighted signals sum to 1. Provider coverage — the fraction of your available assistants that actually returned an answer — scales the result rather than adding a floor. That's a deliberate honesty choice: a brand no assistant names scores near 0, not a flattering 10.
Share of AI Voice
Share of AI Voice is reported alongside the score and is the most competitive of the signals. It answers “of all the brand mentions in these answers, what share were yours?”
A brand can have a modest mention rate but a strong share of voice in a small category, or vice-versa — which is why we report both.
How to read a score
- Direction beats absolute value. The most useful signal is whether the score is climbing after you act.
- It's relative to a prompt set. The score is always measured against a specific set of buyer questions and selected assistants — not a universal ranking in any AI system.
- Comparisons are like-for-like. Our public rankings and leaderboard score every brand in a category the same way, so head-to-heads are fair.
Limitations we're honest about
- AI assistants do not publish rankings, impressions or search volume — every measurement is an estimate built from sampling, not a figure read from an API.
- Answers shift with model updates, phrasing and date. We sample and track over time to smooth this, but a score is a measurement, not a guarantee.
- We measure visibility against your configured prompts and providers. We do not claim to influence or represent any fixed ranking inside ChatGPT, Claude, Gemini or Perplexity.
Versioning
This methodology is versioned. When we change how the score is computed, we bump the version and note it here, so historical comparisons stay interpretable. This is v1.0.
See your score
Run a free scan and get your AI Visibility Score, measured exactly as described above.
Run your free scan