AI share of voice is your share of the brand mentions in AI answers.
AI share of voice is the number of AI answers that name your brand divided by the total brand mentions across a fixed set of brands, on the same buyer questions and engines. The formula is one line. The hard part is choosing which brands go in the denominator, because that choice can move the figure more than a month of work.
What is the formula for calculating AI Share of Voice?
Share of voice is your mentions divided by the mentions of every brand in the frame, yours included, over the same set of answers.
share of voice = answers naming your brand / sum of answers naming each brand in the frame x 100
In the public Tesla example (4 Oct 2026), 25 unbranded buyer questions were asked on each of four engines, 100 answers in all. Tesla was named in 53, Rivian and SunPower in 13 each, Ford in 10 and Enphase Energy in 9. Across those five brands there are 98 mentions, so Tesla's share of voice is 53 / 98 = 54.1%, and Rivian's is 13 / 98 = 13.3%.
Count each brand at most once per answer. Without that rule a single long answer that repeats a name five times outweighs four answers that name it once, and the numerator no longer counts the same thing as the denominator.
Why does the set of brands change the answer?
Add one more brand to the frame and every share shrinks. Put General Motors, named in 7 of the same 100 Tesla answers, into the frame and the total becomes 105: Tesla's share of voice falls from 54.1% to 50.5% without a single answer changing. A vendor that quietly picks a smaller frame reports a bigger number.
So a share of voice figure is only meaningful with its frame named. There are three honest frames:
| Frame | Denominator | Good for |
|---|---|---|
| Tracked brands | Your brand plus the competitors you chose to track | Watching the rivals you already know |
| Every vendor named | Every brand any answer named, tracked or not | Finding rivals you did not know the engines recommend |
| Top 10 vendors named | The ten most-named vendors, plus yours if you rank lower | Comparing brands and categories on one stable scale |
Proofsource prints each share under one of exactly these three names, with the counts behind it, for example “53 of 98 mentions across 5 tracked brands”, so a reader always knows what was divided by what.
How is share of voice different from mention rate and citation share?
They answer three different questions, and they can move in opposite directions. When engines start naming more brands per answer, your mention rate can hold steady while your share of voice falls.
- Mention rate
- Answers that name you / all answers. Are you on the shortlist at all?
- Share of voice
- Your mentions / all brand mentions in the frame. How much of the shortlist is yours?
- Citation share
- Citations to your site / all citations. How much of the evidence behind the answers is yours?
- Citation rate
- Answers that cite your site at least once / all answers. How often does an answer send the reader to you?
The terms are defined in the glossary, and the AI visibility score page shows how the four combine into one number.
How should I define and calculate AI citation share for my category?
Citation share is the citations that point at your domain divided by every citation in the same answers. Count by domain, not by URL, so your pricing page and your blog both count as you.
citation share = citations to your domain / all citations in the answers x 100
It is usually far smaller than share of voice, because answers cite third-party pages: reviews, listicles, forums and videos. In the Tesla example the 100 answers cited 455 different websites, and fewer than 2 in every 100 citations went to tesla.com, even though Tesla was named in 53 of the answers. Being named and being cited are different wins, which is why the citations view ranks every cited domain, not only yours.
Can AI Share of Voice be segmented by topic, market, or customer intent?
Yes, as long as every answer is stored with its question, engine and date, because a segment is just a filter on those records. Group your questions by topic or by buying intent and compute the same formula inside each group.
Segments get small quickly, so give every segment its own interval. A topic with 5 questions on 4 engines is 20 answers, and a rate from 20 answers can sit 20 points either side of the truth. Proofsource breaks results down by engine and by topic, and prints no percentage on fewer than 10 answers. On Custom, questions can also be asked from each market you sell in and compared market by market, switched on at onboarding.
How do AI visibility tools calculate share of voice?
Most tools use some version of the formula above, but not all publish the frame or the weighting. Here is what each says on its own pages:
| Tool | What its page says |
|---|---|
| Semrush | Toolkit: a 0 to 100 benchmark of how often your brand appears in AI answers compared to competitors. Index: 126 million US prompts on ChatGPT, Google AI Mode, Google AI Overviews and Gemini, reported per platform, combining how often a brand appears with how consistently. |
| Peec AI | Visibility = responses mentioning your brand / total responses x 100. Share of voice = your brand's mentions / all brand mentions x 100. |
| Scrunch | Presence = present / total x 100. Share of voice = the share of AI answers that mention you compared to competitors, typically aggregated over 90 days. Citations share of voice = responses citing a source you own / responses citing any source x 100. |
| Otterly.AI | Coverage = the share of AI answers your brand appears in. Share of voice = your slice of every brand mention across the set you track, yours and your competitors'. |
- Semrush KB: AI SEO metrics www.semrush.com/kb/1594-ai-seo-metrics
- Semrush AI Visibility Index: methodology ai-visibility-index.semrush.com/methodology
- Peec AI docs: Visibility docs.peec.ai/metrics/brand-metrics/visibility
- Peec AI docs: Share of Voice docs.peec.ai/metrics/brand-metrics/share-of-voice
- Scrunch: How does Scrunch measure AI share of voice scrunch.com/faqs/how-does-scrunch-measure-ai-share-of-voice
- Scrunch help center: metrics guide helpcenter.scrunchai.com/en/articles/13566677-scrunch-metrics-guide
- Otterly.AI: AI search analytics otterly.ai/features/ai-search-analytics
Read on each vendor's own page on . Where a page gives no formula we say so rather than guess.
How does Proofsource measure share of voice?
- On unbranded buyer questions, asked on ChatGPT, Perplexity, Claude and Google AI Overviews, with the same wording on each engine.
- Every vendor the answers name is extracted, not only the ones you list, so “every vendor named” and “top 10” frames include rivals you did not know about.
- Counts beside every percentage, and a 95% Wilson interval on the rate, so a small move can be told apart from noise.
- Each figure opens to the answers it was computed from, so you can recount it.
The method, including the floors below which a figure is withheld, is on the methodology page.
What people ask about AI share of voice
What is a good AI share of voice?
It depends on how crowded the answers are. With ten vendors in the frame an equal share is 10%, so 20% means you are named twice as often as an average rival. Proofsource's AEO score treats 25% of the top-10 frame as full marks, and says plainly that this anchor is a convention, not a law.
Should share of voice count mentions or answers?
Count each brand at most once per answer. Then a long answer that repeats one brand five times does not outweigh four short answers, and the numerator and the denominator count the same thing. Proofsource counts a brand once per answer.
Should branded questions count toward share of voice?
No. A question that names your brand will mention it whether or not the engine would ever recommend you, so it inflates the figure. Measure share of voice on unbranded buyer questions and report branded questions separately.
Which tools can track brand mentions across ChatGPT, Gemini, and Perplexity?
Several do, including Peec AI, Otterly.AI, Scrunch and Semrush, each on its own engine list. Proofsource tracks ChatGPT, Perplexity, Claude and Google AI Overviews on every plan; Gemini, Google AI Mode, Microsoft Copilot and others are added on the Custom plan.
How often should share of voice be measured?
Often enough that the interval narrows before you act. On the Custom plan Proofsource asks every question on every engine once a day. The free trial runs 2 scans, today and tomorrow, with 25 questions on each engine.
See your share of voice, with its frame named.
See which buyer questions name your brand, who gets named instead, and what to fix. 200 answers. 25 Prompts/Question. Top AI engines. 2 days. Free. No card.
Google AI OverviewsGoogle AI Mode
