AI share of voice is the portion of brand mentions in AI-generated answers that belong to you, compared with the competitors you track. When ChatGPT, Gemini, Perplexity, Claude or Google’s AI Mode name five brands in reply to a buyer’s question, share of voice tells you how often one of those names is yours.

It is one of the most useful numbers in AI search because it puts your visibility in context. A rising mention count looks good until you notice that your main competitor rose faster. This guide explains what AI share of voice means, how to calculate it, how to measure it reliably across engines and prompts, how to set benchmarks from your own data instead of borrowed figures, and what actually grows it.

Key takeaways

  • AI share of voice compares your presence in AI answers with a defined set of competitors. It is a relative metric, so it can fall even when your own mentions hold steady.
  • The simplest version counts mentions. Stronger versions also weigh your position in the answer and the citations that point to your pages.
  • The number is only as good as the prompts and competitors behind it. Use unbranded prompts that real buyers ask, and a competitor list that reflects who AI engines actually name.
  • Measure it per engine, per topic and per region before you roll it up. A single total hides where you win and where you lose.
  • Set benchmarks against your own baseline, your fair share and the leader in each topic. Generic industry targets do not transfer to AI answers.
  • You grow it by becoming the best-supported answer to the questions in your category: pages that match what engines search for, corroboration on the sites they cite and accurate descriptions everywhere.

What is AI share of voice?

Share of voice is an old marketing idea. In advertising it meant your portion of the total ad spend or impressions in a category. In SEO it usually means your portion of the estimated visibility or clicks for a set of keywords. AI share of voice applies the same idea to answers written by AI systems.

The unit is different, though. There is no list of ten links to rank in. An AI answer names a handful of brands, sometimes in a ranked list and sometimes in passing, and links to a few sources. AI share of voice measures how much of that limited space you hold across many answers, compared with the brands you compete with.

How it differs from search share of voice

  • Answers, not rankings. In classic search every page has a position. In an AI answer you are either named or not, and if you are named, you appear somewhere in the text.
  • Few slots. A recommendation answer often names only a handful of brands. Missing from it means missing from the shortlist entirely.
  • Variable output. The same prompt can produce different brands on different runs. Share of voice has to be measured over many samples, not read off one result.
  • Several engines. ChatGPT, Gemini, Claude, Perplexity, Grok and Google’s AI Overviews and AI Mode each draw on different sources. Your share can differ sharply between them.

Why it matters

When someone asks an assistant for the best tool for a job, the answer often is the shortlist. People may click a source or two, or they may simply search for the brands named and ignore the rest. Being named regularly, early and favorably shapes which brands get considered before anyone visits a website.

Share of voice also catches competitive shifts that absolute numbers miss. If a competitor publishes a strong comparison page or earns coverage on a review site that engines trust, their share rises and yours falls, even if nothing on your own site changed. That is a signal worth catching early.

How to calculate AI share of voice

There is no single industry formula, so define yours clearly and keep it fixed. The four versions below answer slightly different questions. Most teams start with the first and add the others as their program matures.

Mention share

Mention share is the basic version. Count the mentions of every tracked brand across the answers in your sample, then divide your mentions by the total.

Mention share = your mentions / (your mentions + mentions of all tracked competitors)

Count a brand once per answer, even if the answer names it three times. Otherwise long answers that repeat a name inflate the result, and your number starts to measure answer length rather than visibility.

Position-weighted share

Being named first in a list of six is worth more than being named sixth. A position-weighted share gives each mention a weight based on where it appears, for example 1 for first, one half for second and one third for third. You then divide your weighted total by the weighted total for all tracked brands. This rewards the brands that engines lead with, which is closer to how readers experience the answer.

Citation share

Citation share looks at sources rather than names. Of all the citations that point to your domain or to a tracked competitor’s domain, what portion point to you? This tells you whose pages engines rely on as evidence. It is a different signal from mentions, because an engine can cite a page without naming its brand, and name a brand without citing it. Our guide to brand mentions vs citations explains why both matter.

Answer presence

Answer presence asks a simpler question: in what share of your tracked prompts do you appear at all? It is not strictly share of voice, because it ignores competitors, but it pairs well with it. A brand can hold a high share in the few prompts where it appears while being absent from most of its category.

A worked example

Suppose you track 40 unbranded prompts in one engine, and you and four competitors are named across the answers. Over a week of daily runs you collect 280 answers. Your brand is named in 70 of them, and the five brands together collect 350 mentions, each counted once per answer. Your mention share is 70 divided by 350, which is 20 percent.

Now suppose that when you are named, you tend to appear fourth or fifth, while the leader is usually named first. Your position-weighted share will be lower than 20 percent, and the leader’s will be higher. That gap is often the most useful finding: engines know you, but they do not lead with you. These numbers are only an illustration. Yours will depend on your category, prompts and competitors.

Tip: Write your formula down, including how you count repeated names, which brands are in the competitor set and which prompts are included. If any of those change, mark the date in your reports so nobody compares numbers that were calculated differently.

How to measure it across engines and prompts

Share of voice depends entirely on the sample behind it. Two teams measuring the same brand can get very different numbers if they use different prompts or competitors. These are the choices that matter most.

Use unbranded prompts that buyers really ask

A prompt that includes your brand name will almost always mention you, which tells you nothing about how engines find you on their own. Build your set from category questions instead: best options for a use case, comparisons between products, how to solve a problem and what a good solution costs. Write them the way customers speak, with the context they add, such as company size, industry or country. The free AI prompt ideas generator can help you draft a first list.

Group prompts into four to seven topics, such as pricing, integrations or specific use cases. Share of voice by topic is far more useful than one blended number, because you will usually lead some topics and trail others.

Pick a realistic competitor set

Your competitor set is the denominator, so it shapes the result. Include the brands AI engines actually name for your prompts, not only the ones your sales team talks about. Run your prompts first and look at who appears. You may find directories, marketplaces or a brand from an adjacent category taking slots you assumed were yours.

Keep the set focused. Adding every company in your industry dilutes everyone’s share and makes changes harder to read. Five to ten real rivals is a sensible range for most brands.

Sample enough runs

AI answers vary between runs, so a single run per prompt is noisy. Daily tracking over at least two weeks shows you the normal spread. That spread matters later: if your share usually moves two or three points from week to week, a two-point change is not news.

Segment before you roll up

Measure per engine first. Each engine has its own index, retrieval habits and preferred sources, and a brand that leads in Perplexity may be missing from Gemini. Then cut the numbers by topic, by region and by language if you sell in several markets. Roll them up for reporting, but make decisions from the segments.

  • By engine: find where you are weakest and look at what that engine cites instead of you.
  • By topic: find the themes where one competitor dominates.
  • By region and language: answers differ by location, and so do the local sources engines trust.
  • By intent: separate early research prompts from comparison and buying prompts, since the second group usually matters more to revenue.

Benchmarks to set for yourself

It is tempting to ask what a good AI share of voice is. There is no honest universal answer. The number depends on how many competitors you track, how crowded your category is, which engines you measure and how your prompts are written. A figure that is strong in one setup can be weak in another, so borrowed benchmarks mislead more than they help.

Set benchmarks from your own data instead. These five work well together.

Your baseline

Record your share per engine and per topic over the first two to four weeks. That is your starting point. Every target after that is a change against the baseline, measured on the same prompts and competitors.

Your fair share

If you track yourself and four competitors, an even split gives each brand 20 percent. That is your fair share, calculated as 100 divided by the number of brands in the set. Being above it in a topic means engines favor you there. Being below it shows where you are underrepresented.

The gap to the leader

In each topic, note who holds the largest share and how far behind you are. Closing a specific gap with a specific competitor is a clearer goal than “grow share of voice,” and it points you at the pages and sources that competitor wins with.

Engine floors

Set a minimum share you want to hold in each engine that matters to your audience. A floor protects you from a blind spot, such as strong results in ChatGPT hiding near-absence in Google’s AI Mode.

Movement targets

Once you know your normal week-to-week swing, set targets in terms of sustained change beyond it. For example, aim to move a weak topic from below your fair share to above it over a quarter. Review the targets each quarter, because markets change and so do the engines.

Tip: Report share of voice alongside your own mention rate. If share of voice falls while your mention rate holds steady, a competitor is gaining. If both fall, you are losing ground directly. The response is different in each case.

How to read changes in share of voice

Because share of voice is relative, it moves for reasons that have nothing to do with your own work. Before you act on a change, work through these questions.

  1. Did the sample change? New prompts, paused prompts or a new competitor change the denominator. Compare only prompts that were answered in both periods.
  2. Is it one engine or all of them? A drop in a single engine often traces back to a change in what that engine cites. A drop everywhere points to something broader, such as a competitor’s new content or coverage.
  3. Which topics moved? Share rarely moves evenly. Find the topic where the change is concentrated.
  4. Who gained? Identify the competitor whose share rose, then look at the sources cited when they are named. The explanation is usually there.
  5. Is it beyond normal noise? If the change is within your usual spread, wait another week before drawing conclusions.

How to grow your AI share of voice

There is no shortcut that reliably adds a brand to AI answers. Engines name the brands they find well described and well supported in the sources they retrieve. Growing your share means becoming that brand for more of the questions in your category.

Start with topics where you are absent

Your biggest gains usually come from prompts where you are not named at all, not from moving up in answers where you already appear. List the prompts where competitors are named and you are not, group them by topic and pick the topic closest to your revenue. Then find out why you are missing. Is there no page that answers the question, or does the page exist but never get retrieved or cited?

Earn presence on the sources engines already cite

Look at the domains cited in the answers where competitors are named. They are often review sites, comparison lists, directories, community forums and industry publications. If engines lean on these for your category, your own site alone will not close the gap. Get listed accurately, update outdated profiles, contribute useful answers in communities where that is welcome, and pitch the authors of lists that leave you out. The earn third-party citations playbook covers this step by step.

Publish honest comparison and alternatives pages

Many of the prompts that decide share of voice are comparisons: one product against another, alternatives to a known brand, or the best option for a particular kind of team. Engines need sources that answer these directly. A fair comparison page that states who each option suits, with specific facts on pricing, features and limits, gives them something to quote. See the comparison pages playbook for a structure that works.

Cover the follow-up questions engines search for

When an engine answers a broad prompt, it often runs several narrower searches behind the scenes about pricing, integrations, setup and drawbacks. This is called query fan-out, and it means your share of voice depends on how well you answer those sub-questions too. Read query fan-out explained for how to find and cover them.

Fix how you are described

Share of voice counts mentions, but the description decides whether a mention helps you. Check the context of your mentions. Is your pricing current, are your features right, are you placed in the correct category? Correct wrong claims at their source, on your own pages and on the third-party sites engines cite. A brand that is described accurately as the best fit for a specific use case tends to be named for that use case.

Keep pages reachable and current

None of this works if engines cannot read your pages. Confirm that the AI crawlers you want are allowed, that key content is in the HTML rather than loaded only by scripts, and that pages covering changing facts show an update date. Our guide to how AI search engines choose sources walks through the signals that tend to matter.

Common mistakes

  • Counting branded prompts. They inflate your share and hide real gaps.
  • Changing the competitor list often. Every change shifts the denominator and breaks your trend line.
  • Reading one blended number. A total can hold steady while you lose a key topic or engine.
  • Chasing daily swings. Judge trends over weeks, not days.
  • Treating all mentions as equal. A passing mention in sixth place is not the same as being recommended first. Track position alongside share.
  • Ignoring sentiment. A high share built on negative or inaccurate descriptions is a liability, not an asset.
  • Measuring without a plan to act. Give each weak topic an owner and a next step.

How aeoora helps

aeoora checks the AI engines daily for your tracked prompts and records which brands each answer names, where they appear and which pages are cited. Competitors are scored on the same answers as you, so the numbers compare directly. Share of voice appears on the Visibility leaderboard and on the topic leaderboard, so you can read it overall and by theme.

  • Prompts and topics: add your own prompts or review suggestions, up to 25 per refresh, and group them into topics. Each prompt shows its query fan-outs, the sub-questions engines search for.
  • Competitors: track 5 competitors per brand on Shine, 10 on Signal and 25 on Scale, with head-to-head comparisons and a breakdown by engine.
  • AI Visibility Score: a 0 to 100 score built from mention rate, citation rate and position, calculated the same way for every competitor. See how the score works.
  • Citations: Competitor Gaps lists the sources cited for competitors but not for you.
  • Alerts: a competitor surge, or a competitor moving above you, shows up in the Daily Pulse and in Slack.
  • Action Center: signals such as a competitor gaining become prioritized actions with tasks, and results are measured at 7, 14 and 30 days.

Read Competitors and share of voice in the docs for how each view works.

Next steps

  1. Draft 20 to 40 unbranded prompts across four to seven topics, then run them in the engines your buyers use.
  2. List the competitors that actually appear, and settle on a set of five to ten.
  3. Record your baseline share per engine and per topic over two weeks, and calculate your fair share.
  4. Pick the topic with the largest gap to the leader and work through the growth steps above.

To see where you stand today, request a free AI visibility report, which checks 25 buyer questions and emails you a PDF. When you want daily tracking, compare plans on the pricing page or start a 2-day free trial. For related reading, see how to measure AI search visibility, the prompts and topics guide and the beat a competitor in AI answers playbook.

Written by

AI search research

Share this article