Visibility Score
The visibility score is a number from 0 to 100 for one audit. It is a weighted average of four metrics, multiplied by 0.2 + 0.8 × coverage, where coverage is the market-weighted share of the platforms in your audit that mentioned you at least once. It is our model of visibility, built from measured answers. It is not a rank any AI platform publishes.
Formula
overall = round(average × multiplier), where the average weights the four metrics below and the multiplier is 0.2 + 0.8 × coverage.
- Mention Rate (40%) — the share of answers that name your brand. Each prompt asked on each platform is one answer: 20 prompts on 6 platforms is 120 answers, and if 18 of them name you, your mention rate is 15%.
- Accuracy (30%) — starts at 100 and drops with each statement an answer gets wrong against your brand facts, weighted by how serious the worst error is and how often errors occur. Two checks (acquisition claims and staff counts) run even with no facts on file, so it can drop below 100 on a brand whose audit shows accuracy as "Not checked".
- Sentiment (15%) — the tone of the answers, from 0 (negative) through 50 (neutral) to 100 (positive), averaged over every answer.
- Citation Rate (15%) — for each answer, the share of its links that point to your site, averaged over every answer; an answer with no links counts as 0. Every platform can cite sources when it searches, ChatGPT and Claude included; an answer given from memory cites nothing.
Coverage
That average is then multiplied by your platform coverage: the market-weighted share of the platforms in your audit that mentioned your brand at least once. The weights are ChatGPT 36%, Google AI Overviews 18%, Perplexity 18%, Google AI Mode 10%, Gemini 9% and Claude 9%. The multiplier is 0.2 + 0.8 × coverage:
- Mentioned on no platform: your score is 0, whatever your accuracy and sentiment. There is nothing for them to describe.
- Mentioned on every platform: the multiplier is 1 and you keep the full average.
- In between: mentioned only on Perplexity in a 6-platform audit, your coverage is 18% and the multiplier 0.34, so a weighted average of 50 becomes 17.
If Google shows no AI Overview for a query, that query is skipped, not counted as a miss. The per-platform scores use the same four metrics without the coverage multiplier, so a platform that never mentioned you can still show a score from accuracy and sentiment. Read its mention rate first.
Worked example
Mention Rate 30, Accuracy 100, Sentiment 60 and Citation Rate 10 give a weighted average of 52.5. If only ChatGPT and Perplexity named the brand in a 6-platform audit, coverage is 54%, the multiplier is 0.63, and the score is 33.
Metric details
| Metric | Field | Unit | Computed from |
|---|---|---|---|
| Mention Rate | mentionRate | 0–100 | Answers that name the brand ÷ measured answers. A mention matches the brand name, its variants and possessives, or the domain. |
| Accuracy | accuracyRate | 0–100 | The accuracy formula, version v3-2026-10-24. See Hallucination detection. |
| Sentiment | sentimentScore | 0–100 | Each answer's tone from −1 to 1, from positive and negative word counts with words near the mention counting double, averaged and mapped to 0–100. |
| Citation Rate | citationRate | 0–100 | Per answer, own-site links ÷ all links; an answer with no links is 0. Averaged over measured answers. Owned URLs count as own site. |
Per-platform and per-prompt scores
- Same four metrics and weights, from that platform's or prompt's answers only.
- No coverage multiplier. A platform that never named the brand can still score above 0 from accuracy and sentiment.
- Per-platform accuracy needs at least 5 answers from that platform. Below that the audit-wide accuracy is used.
- Per-prompt accuracy has no minimum.
Versions and comparability
Platform weights belong to a weight set, currently v3-2026-10, stored on each audit. Scores from different weight sets are not compared; the audit page then says "No comparison: your last audit was scored with different platform weights." The same holds when brand mentions were counted differently (another mention matcher version, or other brand aliases): "No comparison: your last audit counted brand mentions differently." The dashboard home, the scheduled-audit email and the before/after of a finished fix apply the same rule, and the home trend marks where the scoring changed. Accuracy carries its own formula version for the same reason.
Display
The dashboard colours a score in four bands, with breaks at 40, 60 and 80. The bands are display only; nothing in the product treats 59 and 60 differently.
What the score does not measure
- Position or rank. No engine publishes one, so the score reports presence, accuracy, tone and citations.
- Retrieval on every answer. ChatGPT, Claude and Gemini decide per prompt whether to search, so some answers come from what the model already knows. Each answer is labelled as searched or from memory, so you can tell the two apart.
- Every query a buyer runs. Platforms rewrite questions before they search. Your prompts are a sample, not the exact queries.
- A shared unit across platforms. Platforms cite in different ways, which is why weights and coverage are our model, stated here, rather than a measured constant.
Reading a confusing number on your own dashboard? Don't understand your visibility score? walks through it. Score not moving? Score not improving after your changes.