AI Share of Voice: How to Calculate & Improve It
How to calculate AI share of voice across ChatGPT, Claude, Perplexity and Gemini: formulas, worked examples, a spreadsheet template and what a good SOV is.
AI share of voice is the percentage of brand mentions in AI-generated answers that belong to your brand, compared with your competitors, across a fixed set of prompts. To calculate it, divide your brand's mentions by the total mentions of your brand plus your tracked competitors, then multiply by 100. A brand named in 30 of 100 total competitor mentions has a 30% AI share of voice. Because AI answers vary from run to run, you need repeated sampling across ChatGPT, Claude, Perplexity and Gemini for the number to mean anything.
What is share of voice?
Share of voice (SOV) is a classic marketing metric that measures how much of the conversation in a market belongs to you compared with your competitors. The definition stays the same across channels; only the thing being counted changes:
- Advertising: your ad spend or impressions as a share of total category ad spend or impressions.
- SEO: your share of estimated organic clicks or visibility across a tracked keyword set.
- Social media and PR: your share of brand mentions, conversations or media coverage in your category.
- Paid search: your impression share for the keywords you bid on.
In every case, SOV answers the same question: when people encounter your category, how often do they encounter you instead of someone else?
What is AI share of voice?
AI share of voice applies the same idea to answers from AI assistants. When someone asks ChatGPT, Claude, Perplexity or Gemini "What's the best project management tool for a small agency?", the answer usually names a handful of brands. AI SOV measures how much of that shortlist is yours.
It matters because AI answers compress the market. A Google results page shows ten links plus ads; an AI answer might name three to five brands, and a buyer may never look further. If competitors are named consistently and you're not, buyers who research in AI assistants hear about them instead of you, and classic SEO rank trackers won't show it.
AI SOV is one part of broader AI visibility. If you're new to the topic, start with our AI brand visibility guide for the bigger picture.
How to calculate share of voice: the formulas
There's no single official formula for AI share of voice. Here are the four most useful versions. All numbers below are illustrative examples, not benchmarks.
1. Mention-based SOV (the standard formula)
AI SOV = (your brand mentions / total mentions of you + tracked competitors) x 100
Example: You track your brand and four competitors across 25 prompts, each run once on ChatGPT, Claude, Perplexity and Gemini (100 answers). One run each keeps the math simple; in practice, repeat each prompt 3-5 times. Across those answers, the five brands are mentioned 200 times in total. Your brand is mentioned 40 times.
AI SOV = 40 / 200 x 100 = 20%
This is the version most people mean by "AI share of voice." It's simple, easy to explain, and works well for tracking trends.
2. Position-weighted SOV
Being named first in an answer is worth more than being the fifth name in a list. Weighted SOV gives each mention points based on position. One simple scheme:
| Position in answer | Points |
|---|---|
| 1st brand mentioned | 5 |
| 2nd | 4 |
| 3rd | 3 |
| 4th | 2 |
| 5th or later | 1 |
Weighted SOV = (your total points / total points of all tracked brands) x 100
Example: Your 40 mentions earn 110 points, and all tracked brands combined earn 550 points. Your weighted SOV is 110 / 550 = 20%. If your mentions were mostly first-place, your weighted SOV would come out higher than your mention-based SOV, and vice versa if you're usually listed last. Comparing the two tells you whether you're a top recommendation or an afterthought.
3. Per-platform SOV
Calculate SOV separately for each AI assistant, because they often disagree. An example result:
| Platform | Your mentions | Total tracked mentions | SOV |
|---|---|---|---|
| ChatGPT | 12 | 50 | 24% |
| Claude | 9 | 48 | 19% |
| Perplexity | 14 | 52 | 27% |
| Gemini | 5 | 50 | 10% |
| All platforms | 40 | 200 | 20% |
In this example, the blended 20% hides a weak spot on Gemini, which is useful to know because each assistant can draw on different sources and may respond to different fixes.
4. Citation share
Some AI tools show sources: Perplexity includes numbered citations in its answers, and ChatGPT search includes links to the pages it used. Citation share measures how often your URLs are cited compared with competitors' URLs:
Citation share = (citations to your domain / total citations to tracked domains) x 100
Citation share and mention share are related but different. You can be mentioned without your site being cited (because a third-party review was the source), and your page can be cited without your brand being recommended. Tracking both shows whether your own content or other people's content is driving your visibility.
How to calculate AI share of voice manually (step by step)
You can measure AI SOV with nothing more than a spreadsheet and some patience.
-
Pick your competitors. Choose 3-6 brands that buyers actually compare you with. The competitor set defines the denominator, so keep it fixed once you start.
-
Build a prompt set. Write 20-30 prompts buyers would realistically type. Mix category, comparison and problem-based questions:
What is the best [category] for [use case]? Top [category] tools for [audience] [Competitor] alternatives [Your brand] vs [competitor]: which is better? How do I [problem your product solves]? -
Run every prompt on every platform. Use ChatGPT, Claude, Perplexity and Gemini. Use a logged-out or fresh session where possible to reduce personalization.
-
Repeat each prompt several times. AI answers vary from run to run, so a single run is a sample, not a measurement. Three to five runs per prompt per platform is a reasonable start.
-
Record every brand mentioned and its position in a spreadsheet (template below).
-
Count and calculate. Sum mentions per brand, then apply the formulas above, overall and per platform.
-
Repeat on a schedule. Monthly works for most brands, plus after big launches, PR pushes or model updates. Watch the trend, not individual runs.
Spreadsheet template
| Date | Platform | Prompt | Run # | Brands mentioned (in order) | Your position | Your brand mentioned? | Sources cited | Sentiment |
|---|---|---|---|---|---|---|---|---|
| 2026-10-01 | ChatGPT | Best CRM for small agencies | 1 | Brand A, You, Brand B | 2 | Yes | — | Positive |
| 2026-10-01 | Perplexity | Best CRM for small agencies | 1 | Brand B, Brand C | — | No | g2.com, brandb.com | — |
Then add a summary tab with one row per brand: total mentions, total position points, mentions per platform, and the resulting SOV percentages.
For more detail on prompt design and logging, see how to track ChatGPT brand mentions and our full guide on how to track brand visibility in AI platforms.
Limitations to keep in mind
- Variability: the same prompt can return different brands on different runs, days, locations and model versions.
- Personalization: memory and chat history can shape answers for logged-in users (see OpenAI's memory FAQ).
- Prompt selection bias: your SOV only reflects the prompts you chose. Different prompts, different number.
- Competitor set: adding or dropping a competitor changes everyone's percentage.
None of this makes AI SOV useless. It means you should treat it as a trend indicator measured consistently, not a precise market fact.
What is a good share of voice percentage?
There's no universal "good" share of voice percentage. It depends heavily on how many competitors you track and how big you are in the market.
Start with the even-split baseline. If you track yourself plus four competitors (five brands), an equal share would be 20%. Above that, you're getting more than your "fair" share of AI mentions; below it, less. With three brands the baseline is about 33%; with ten brands it's 10%.
Compare against market share. A widely cited marketing rule of thumb, often called excess share of voice, says brands whose share of voice is higher than their market share tend to grow. Les Binet and Peter Field's research for LinkedIn's B2B Institute, The 5 Principles of Growth in B2B Marketing, found that pattern in B2B too. It comes from advertising research and is a rule of thumb, not a law, but in our view it's a useful lens for AI answers. For example, if you hold roughly 15% of your market and AI assistants give you 25% of mentions, AI is amplifying you. If they give you 5%, competitors are capturing buyers who start their research in AI tools.
Look at the trend and the leaders. A rising SOV month over month matters more than any single figure. It also helps to know the gap to the category leader and whether that gap is shrinking.
What about the extremes?
- 100% share of voice means you're the only tracked brand mentioned across your prompt set. That's rare in a competitive category, and it often means your prompts are too branded.
- 50% share of voice means half of all tracked mentions are yours, which in a multi-competitor category usually signals clear category leadership for those prompts.
How to improve your AI share of voice
AI SOV improves when AI models see more, and more consistent, evidence that your brand belongs in the answer.
- Find the prompts you lose. Filter your spreadsheet for prompts where competitors are named and you're not. Those are your priorities.
- Check which sources get cited. For prompts you lose, note the URLs Perplexity, ChatGPT search and Gemini cite. Those roundups, review sites and forums are where you need to appear.
- Earn third-party mentions. Get included in "best [category]" lists, collect reviews on platforms relevant to your industry, and take part honestly in community discussions.
- Publish comparison and alternatives pages. Honest "[You] vs [Competitor]" and "[Competitor] alternatives" pages give retrieval-based AI tools quotable content for exactly the prompts you track.
- Write answer-first. Lead pages with clear definitions, step lists and comparison tables that models can lift.
- Keep your entity consistent. Describe your brand the same way on your site and every profile and directory.
- Make sure AI crawlers can reach you. Check robots.txt and bot protection, and serve key content in the HTML. Our AI search engine optimization playbook covers the technical side.
- Re-measure with the same prompt set. Keep prompts and competitors fixed so you can see whether changes worked.
Share of voice tools
You have three practical options for measuring AI share of voice:
| Approach | How it works | Pros | Cons |
|---|---|---|---|
| Manual spreadsheet | Run prompts yourself and log results | Free, full control, you see every answer | Slow, hard to repeat at scale, easy to introduce inconsistency |
| AI visibility tools | Software queries AI assistants with prompts and reports mentions, rankings and SOV | Fast, repeatable, covers multiple platforms | Paid; prompt sets may be generated for you rather than chosen |
| Classic SOV tools (SEO, social, PR) | Measure share of organic search, social mentions or media coverage | Mature, good for traditional channels | Core reports don't measure AI answers (Google Alerts only watches Google Search results); some vendors now sell separate AI add-ons |
Classic SEO and social listening tools are still useful for their own channels, but their core reports don't tell you what ChatGPT or Claude say about your category. For that you need the manual method, a dedicated tool, or one of the AI add-ons some established vendors now sell. As of October 2026, for example, Semrush's AI Visibility Toolkit runs prompt tracking on ChatGPT and Google AI Mode, and Meltwater's GenAI Lens reports AI share of voice across several assistants. Coverage differs a lot between tools, so compare platform lists before buying. We compare the options in our roundup of the best AI visibility tools.
AI Brand Track is one of those tools. You enter your website URL, it extracts your brand, industry and keywords, finds your competitors, then queries ChatGPT, Claude, Perplexity and Gemini with industry prompts. In about a minute you get a 0-100 visibility score, competitor rankings, which prompts mention you, sentiment, share of voice and recommendations. See how it works on the AI rank tracker page, or start a 14-day trial for $9 (1 full analysis plus weekly checks of 1 brand) to get your baseline before building a manual process.
FAQ
What does AI share of voice mean?
AI share of voice is the percentage of brand mentions in AI assistant answers that belong to your brand, relative to the competitors you track. It's measured across a fixed set of prompts on tools like ChatGPT, Claude, Perplexity and Gemini. It shows how often AI recommends you compared with alternatives.
How do I calculate my share of voice?
Divide your brand's mentions by the total mentions of your brand plus your tracked competitors, then multiply by 100. For AI share of voice, count mentions across a consistent prompt set, run each prompt several times on each platform, and average the results. You can also weight mentions by position or calculate SOV per platform.
What is a good share of voice percentage?
It depends on how many competitors you track and your market position. A useful baseline is an even split: with five tracked brands, 20% is your "fair" share. A widely cited marketing rule of thumb, backed by Les Binet and Peter Field's research, is that brands with SOV above their market share tend to grow.
What does 50% share of voice mean?
A 50% share of voice means half of all tracked brand mentions in your prompt set are yours. In a category with several competitors, that usually means you're the dominant recommendation for those prompts. Check that your prompts aren't overly branded before celebrating.
What does 100% share of voice mean?
A 100% share of voice means your brand is the only tracked brand mentioned across your prompts. In competitive categories that's very unusual. It typically means the prompt set is too narrow or includes your brand name, so broaden it to category and problem-based questions.
What does share of voice tell you?
Share of voice tells you how visible your brand is compared with competitors in a given channel. In AI search, it shows whether assistants recommend you or your competitors when buyers ask category questions. Tracked over time, it shows whether your content, PR and reputation work is shifting AI answers in your favor.
How to improve share of voice?
Find the prompts where competitors appear and you don't, then look at which sources AI tools cite for them. Earn mentions in those sources, publish honest comparison and alternatives pages, write answer-first content, and keep your brand description consistent across the web. Re-measure with the same prompts to confirm progress.
What does SOV stand for in marketing?
SOV stands for share of voice. It measures your brand's share of visibility or conversation in a market compared with competitors, whether that's ad spend, organic search, social mentions, media coverage or, more recently, mentions in AI-generated answers.