AI visibility measurement
AI share of voice
AI share of voice estimates how much of a defined answer set your brand occupies compared with competitors. The number can be useful, but only when the prompt set, answer surfaces, market, scoring rule, and denominator are visible.
Updated July 14, 2026
In short
AI share of voice is your brand's share of mentions, citations, or weighted visibility inside a defined set of AI answers. To calculate it honestly, define the buyer questions, assistants, market, competitors, time window, and counting rule first. A 30 percent score from 20 branded prompts is not comparable with 30 percent from 2,000 non-branded category questions.
The differences at a glance
| Aspect | Mention share | Recommendation share |
|---|---|---|
| What counts | The brand appears somewhere in the answer | The answer gives the brand a meaningful reason to be considered |
| What it is good for | Basic presence and competitive coverage | Buyer-facing persuasion and fit |
| What it can hide | Weak mentions, stale descriptions, and stronger competitor recommendations | Broad awareness that does not yet carry a recommendation |
| Best next question | Where are we missing? | Why did the buyer get a stronger reason to choose someone else? |
The denominator is the metric
Share of voice sounds universal, but the score is produced by a measurement design. Twenty friendly branded prompts, fifty buyer comparisons, and a large search-backed prompt index answer different questions. A trustworthy report names the eligible prompts, surfaces, countries or languages, competitor set, cadence, failed reads, and rule used to count each answer.
Three different numbers are often called AI share of voice
Tools commonly use one of three approaches. Answer share asks how many eligible answers mention the brand. Competitive mention share divides one brand's appearances by all monitored brand appearances. Weighted share adds position, citation, recommendation strength, or modeled demand. None is automatically wrong, but they should not be compared as if they were the same metric.
- Answer share: answers mentioning your brand divided by eligible answers.
- Competitive mention share: your brand mentions divided by mentions across the chosen peer set.
- Weighted share: visibility adjusted by factors such as answer position, citations, recommendation strength, or modeled prompt demand.
Why two AI visibility tools can disagree
One tool may use prompts you supplied and run them daily. Another may model a much larger search-backed question set. One may count any mention, while another weighs position or citations. They may query different answer surfaces, countries, or product modes. The useful response is not to average the scores; it is to inspect which measurement design matches the buying questions you care about.
Measure ChatGPT separately when it matters
A blended score can hide a real platform gap. If buyers in your category use ChatGPT heavily, keep a ChatGPT share alongside the all-surface total. Do the same for Perplexity, Gemini, Claude, Google AI Overviews, or AI Mode when their retrieval behavior and audience matter enough to change the decision.
Recommendation share asks the harder question
A brand can win mention share and still lose the sale. It may appear in many answers as an afterthought while a competitor receives the confident fit statement. Signalbat keeps recommendation share beside presence: was the brand merely named, placed on a shortlist, recommended for the use case, preferred, or weakened by a caveat?
A practical calculation
Suppose you monitor 30 non-branded buyer questions in ChatGPT. Your brand appears in 12 answers, so its answer share is 40 percent for that set. It receives a meaningful recommendation in 6, so recommendation share is 20 percent. A competitor appears in 10 but is recommended in 8. You have the larger footprint; the competitor has the stronger buying story.
Turn the score back into evidence
When share moves, inspect the answers underneath it. Which prompt family changed? Did a new competitor enter? Did a citation start repeating? Is the brand present but described for the wrong use case? The metric earns its place when it leads to a clearer page, stronger proof, a fair comparison, or better third-party context.
Do not report false precision
AI answers vary, prompt samples are finite, and failed reads happen. Report the sample size and time window, keep raw answers available, and avoid presenting a tiny movement as market truth. Share of voice is a controlled read of selected answer surfaces, not a census of every private AI conversation.
How to measure AI share of voice
- 01
Define the question set
Lock the buyer questions, prompt families, markets, languages, answer surfaces, and competitor set included in the read.
- 02
Choose the counting rule
Decide whether the metric counts mentions, citations, recommendations, position, or a documented weighted combination.
- 03
Preserve the answers
Keep raw answer wording, timestamps, sources, failed reads, and model or surface metadata behind every rollup.
- 04
Explain the movement
Break the change back down by prompt family, competitor, claim, and source before deciding what to improve.
What an honest report should disclose
- Eligible prompts
- Answer surface
- Market and language
- Competitor set
- Mention rule
- Recommendation threshold
- Raw answers
- Sample size
- Change by prompt family
AI share of voice, answered
- What is AI share of voice?
- AI share of voice is a brand's share of mentions, citations, recommendations, or weighted visibility inside a defined set of AI-generated answers. The definition is incomplete unless the report also names the prompts, surfaces, market, competitors, time window, and counting rule.
- How do you calculate AI share of voice?
- A simple version divides eligible answers that mention your brand by all eligible answers. A competitive version divides your brand's mentions by mentions across the monitored peer set. Some tools use weighted formulas. Pick one method, document it, and keep it stable enough to compare over time.
- What is a good AI share of voice?
- There is no universal good percentage. A useful benchmark is your own stable baseline, the competitors in the same question set, and whether share is improving on high-intent questions. Ten percent on the questions that drive consideration can matter more than fifty percent on friendly branded prompts.
- Why do AI visibility tools report different share-of-voice scores?
- They may use different prompt sets, assistants, countries, schedules, competitor lists, and counting rules. One may count mentions, another position or citations, and another modeled demand. Compare the methodology before comparing the percentages.
- Is AI share of voice the same as market share?
- No. It measures presence inside a controlled answer set, not revenue, customers, or total market demand. Treat it as a directional visibility metric that helps locate buyer questions and evidence worth improving.
- Can I measure ChatGPT share of voice separately?
- Yes. Filter the read to ChatGPT and keep the same prompts, market, competitor set, and scoring rule over time. A platform-specific view is often more useful than a blended total when one assistant matters disproportionately to your buyers.
How current tools define the category
- Ahrefs: Brand Radar methodology Ahrefs explains that its modeled prompts, estimated impressions, and AI share of voice indicate potential visibility rather than actual audience reach.
- Profound: Answer Engine Insights Profound documents daily answer-engine monitoring, visibility and share-of-voice metrics, citations, prompt controls, exports, and enterprise security features.
- Semrush: AI Visibility Toolkit Semrush documents AI visibility, competitor, prompt, perception, share-of-voice, site-audit, export, and reporting workflows inside its wider marketing platform.
- Peec AI: AI search analytics for marketing teams Peec AI's current product page documents prompt tracking, visibility, position, sentiment, source discovery, exports, and API access.
- OtterlyAI: AI search monitoring OtterlyAI documents daily prompt checks, brand mentions, share of voice, website citations, position, sentiment, and reporting across major AI search surfaces.