What is AI share of voice?

How AI share of voice is calculated

The basic formula is: your brand's mentions divided by the mentions of every tracked brand, times 100. Peec AI and Profound both document it this way in their help docs (see Sources). It is always relative: the number only means something next to the competitors and prompts it was measured against. The worked example below uses made-up numbers for one engine and 10 answers.

Share of voice is not the same as visibility or citation share

Visibility (Peec and Profound call it visibility or Visibility Score) is the share of answers that mention you, so it can rise while share of voice falls if competitors are rising faster. Share of voice splits the mentions between the brands you track and always adds up to 100% across them. Citation share counts the source links an answer shows and asks how many point to your domain, which measures whether your pages are used as evidence rather than whether your brand is named.

Why two tools report different numbers for the same brand

Tools agree on the idea but weight it differently. Peec AI and Profound count mentions across the answers to your tracked prompts. Ahrefs Brand Radar weights each prompt by an estimated demand figure it calls AI-adjusted volume: the Google search volume of the keyword behind the prompt multiplied by a platform ratio, currently 0.115 for ChatGPT, 0.011 for Gemini and 0.01 for Perplexity (Ahrefs help center, updated 3 September 2026). So a mention in a popular prompt counts for more than one in a rare prompt. Semrush says its Brand Performance report combines how often a brand is mentioned with how high it appears in the answer, and that Enterprise AIO also factors in topic search volume for ChatGPT. Add different prompt sets, competitor lists and date ranges, and the same brand can honestly score 20% in one tool and 35% in another. Compare a tool with itself over time, not with another tool.

Why AI share of voice moves between checks

AI answers change from run to run. In SparkToro's study with Gumshoe (published 28 January 2026), 600 volunteers ran 12 prompts 2,961 times through ChatGPT, Claude and Google's AI, and fewer than 1 in 100 runs returned the same list of brands; the same list in the same order came up about once in 1,000 runs. The same study found that how often a brand appears across many runs is far steadier than its position in the list. A follow-up by Mike Sonders on Search Engine Land (10 February 2026) ran 12 B2B software prompts 100 times each in ChatGPT: about 44 brands appeared across 100 answers, but only around 5 showed up in 80% or more of them. Share of voice built from a single run per prompt is mostly noise, so look at it over weeks and many prompts.

How to read and report it

Always state the scope with the number: which prompts, which engines, which competitors and which period. Ahrefs makes the same point in its 22 October 2026 guide, where "Our AI SOV is 24% across 500 tracked prompts in ChatGPT and Gemini, against three named competitors, during August" is the version that tells you what the number refers to. Break it down by engine before you trust the blended figure: in Ahrefs' example, 28% on ChatGPT and 11% on Gemini combine to about 26.5% because ChatGPT carries far more estimated impressions, which hides the weak engine. Keep the competitor list fixed between reports, because adding or dropping a rival changes everyone's share. If you pull data through an API, Peec's docs warn that filtering to your own brand before aggregating removes competitors from the denominator and makes share of voice read 100%.

How to raise your AI share of voice

Share of voice goes up when answers name you more often than the brands you track, so the work is the same as getting recommended in the first place: be clearly described on your own pages, be present on the third-party pages AI engines retrieve, and pick prompts where your product is a genuine answer. Narrow categories are easier to win: in the Search Engine Land data, 21% of brands mentioned in niche software categories reached 80%+ visibility against 7% in crowded ones.

Read next: How to get your brand recommended by ChatGPT · How ChatGPT search chooses its sources · AI mentions vs citations · Free AI visibility check.

Tools that report AI share of voice

7 of the 32 tools we track name it in their feature lists, based on each vendor's own site:

• Profound: Answer Engine Insights (citations, sentiment, rank, share of voice). Entry price: Free trial; Enterprise custom. Checked 26 October 2026.
• Peec AI: Visibility, position, sentiment, share of voice. Entry price: €85/mo (unverified). Checked 22 October 2026.
• HubSpot AI Search Grader: Score /100 across sentiment, presence, recognition, share of voice, market position. Entry price: Free. Checked 29 October 2026.
• Contentstack Canoe: Share of voice and rank against up to ten competitors. Entry price: Free; $279/mo. Checked 29 October 2026.
• Temso: Competitor share-of-voice comparison. Entry price: $199/mo (agency Growth, billed yearly). Checked 24 October 2026.
• Serpstat: AI Visibility: brand mentions, sentiment, share of voice and competitor comparison. Entry price: $100/mo. Checked 26 October 2026.
• LLMrefs: Keyword-based tracking: auto-generated fan-out prompts per keyword, aggregated into share of voice and average position. Entry price: $79/mo. Checked 28 October 2026.

Other trackers may show the same metric under another name, such as competitor share or visibility. Compare tools side by side or see GEO tool pricing.

FAQ