Measurement & Tools

July 7, 2026 · Kollavo Team

Updated 13 November 2026

Share of Voice in AI Search: Measuring Your Brand vs Competitors

Share of voice is your share of all brand mentions in AI answers. Learn the formula, how to measure it across ChatGPT, Perplexity and Gemini, and how to act on the gaps.

Share of Voice in AI Search: Measuring Your Brand vs Competitors

Answer: Share of voice (SoV) in AI search is your brand's share of all brand mentions across a set of AI-generated answers. Calculate it as (your mentions ÷ total mentions of all brands) × 100 for the same fixed prompt set, measured across ChatGPT, Perplexity, Claude, and Gemini. Unlike citation rate, which measures your absolute presence, share of voice measures your competitive presence — whether you're winning or losing the answer against rivals.

Citation rate tells you how often AI mentions you. Share of voice tells you whether that's good. A 25% citation rate feels like progress until you learn the category leader sits at 65% — you don't have momentum, you have a gap. This guide shows how to measure AI share of voice and turn it into a competitive roadmap.

Why Share of Voice Beats Citation Rate for Strategy

Citation rate is an inward number: how visible am I? Share of voice is a relative number: how visible am I compared to the brands I'm fighting? In generative search, that distinction is everything, because AI answers are zero-sum in a way blue links never were.

When Google returned ten results, you and three competitors could all rank on page one. When ChatGPT names "the top three tools," being fourth means being invisible. The answer has finite slots, and share of voice measures how many of them you own. That's why executives respond to SoV: it maps directly to competitive position and it's trivially easy to communicate — "we hold 20% of AI answer share; the leader holds 45%."

The Share of Voice Formula

Share of voice = (your brand mentions ÷ total brand mentions) × 100

The "total" is the sum of mentions for every brand you track — you plus your named competitor set — across the same prompt set and platforms. Two refinements make it more useful:

  • Weighted SoV — count a linked citation as worth more than an unlinked mention (for example, 2× or 3×), since citations carry more authority and drive traffic. This rewards quality of presence, not just quantity.
  • Position-weighted SoV — give more weight to being named first or "recommended" than to appearing in a long list. Reflects that prominence, not mere appearance, wins the customer.

How to Measure AI Share of Voice

Step 1 — Define your competitive set

List the 3–8 brands you genuinely compete with for the same answers. Too few and SoV is meaningless; too many and it's diluted. Use the competitors that actually show up in your category's AI answers — the guide below is your customers' questions, not your internal org chart.

Step 2 — Reuse your citation-rate prompt set

SoV runs on the same fixed prompt set you use for citation rate. If you already log which competitors appear on each prompt (you should), you already have the raw data — SoV is just a different calculation over it. Our guide to LLM citation tracking covers how to build and run that prompt set, and the tools that automate it. To compare the trackers that report share of voice, priced per tracked AI answer, see our AI share of voice tools guide. For choosing prompts, engines and the number of runs per check, see our guide to AI prompt tracking.

Step 3 — Count every brand's mentions

For each prompt × platform, record every tracked brand that appears, not just yours. This is the step teams skip, and it's the whole point: without competitor counts there is no share to compute.

Step 4 — Compute overall and segmented SoV

Calculate total SoV, then slice it:

SegmentWhat it reveals
By platformWhere you win (e.g., Perplexity) vs. lose (e.g., Gemini)
By funnel stageWhether you own awareness but lose "best X" decisions
By topic clusterWhich subjects you dominate and which competitors own

Reading the Results: Where SoV Points You

Share of voice isn't a vanity chart — it's a targeting system. The most valuable output is the gap analysis: the specific prompts where a competitor is cited and you are absent. Investigate which competitors AI assistants cite most, then use AI citation gap analysis to shortlist gaps for live verification. Our guide on how to get cited by AI covers how to close those gaps.

  • A competitor owns a topic cluster you care about → that cluster is your content priority. Study what they publish that gets cited, then produce something more useful and more extractable for those exact questions.
  • You're mentioned but never "recommended" → you have presence without prominence. Work on positioning: comparison content, proof, and the authority signals that move you from "an option" to "the pick." Our guide on how to get AI mentions shows how the number of cited pages naming you, and your position on them, drive recommendations.
  • Your SoV is high on one platform, low on another → engines weight sources differently. Reverse-engineer what the winning platform rewards and apply it where you're weak.
  • Overall SoV rising while citation rate is flat → competitors are losing ground faster than you're gaining, which is still a competitive win. Context is everything.

Share of Voice Over Time

A single SoV reading is a snapshot; the trend is the story. Track SoV on the same cadence as citation rate — weekly or biweekly — and watch the lines for you and each competitor together. The most actionable pattern is movement: a competitor whose SoV is climbing fast is executing a strategy worth studying, and a cluster where your SoV is sliding is an early warning to defend before the loss compounds. Use LLM citation decay to distinguish a sustained citation loss from normal variation before choosing a refresh.

Annotate your chart with model updates and your own content launches. When SoV jumps two weeks after you publish a comparison guide, you have a repeatable play. When it drops after a competitor's campaign, you know where to respond.

Common Share of Voice Mistakes

Cherry-picking competitors. Choosing weak competitors inflates your SoV and teaches you nothing. Track the brands that actually win answers in your category, even the intimidating ones.

Ignoring prominence. Treating a first-place recommendation the same as a mention buried in a list overstates weak presence. Use position or link weighting so the number reflects real competitive strength.

Measuring the wrong prompts. SoV on questions no customer asks is a distraction. Anchor the prompt set in real buying-intent language.

One and done. SoV without a trend can't reveal momentum, and momentum is the most decision-useful thing it offers.

What is AI citation share?

AI citation share is the percentage of all source citations in AI answers, for a fixed set of prompts, that point to your brand or your pages. It is a close relative of share of voice: share of voice usually counts every brand mention, linked or not, while AI citation share counts only the cited sources (the links and references an engine shows alongside its answer).

To calculate it, run the same prompt set across the engines you track, count every citation each answer shows, and divide the citations that point to your domain (or to pages that mention you) by the total. Track it on the same cadence as share of voice so the two trend lines can be compared.

Use both together: share of voice shows whether AI talks about you; AI citation share shows whether AI relies on your content or the publications that cover you as evidence.

Read the full guide to AI citation share.

Frequently Asked Questions

What is AI citation share, and how is it different from share of voice?

AI citation share is the portion of all citations in AI answers for your tracked prompts that point to your brand or pages. Share of voice measures your share of all brand mentions in those answers. Citation share is narrower: it only counts sources the engine actually cites.

How is share of voice different from citation rate?

Citation rate measures how often you appear (your absolute presence). Share of voice measures your appearances as a percentage of all tracked brands' appearances for the same prompts (your relative presence). Citation rate tells you if you're visible; share of voice tells you if you're winning.

How many competitors should I track for share of voice?

Three to eight — the brands that genuinely compete for the same AI answers in your category. Too few makes SoV meaningless; too many dilutes it. Base the list on which brands actually appear in your category's answers, not on your internal competitive assumptions.

Should linked citations count more than mentions in SoV?

For a truer picture, yes. Weighting linked citations higher (for example 2–3×) than unlinked mentions rewards quality of presence, since citations carry more authority and drive referral traffic. Report both weighted and unweighted if you want to separate reach from impact.

How often should I measure AI share of voice?

On the same weekly or biweekly cadence as citation rate, using the same fixed prompt set. SoV's biggest insight is momentum — which brands are gaining or losing share — and momentum is only visible in a trend line, not a single snapshot.

What is citation share of voice and how is it different from mention rate?

Citation share of voice is your brand's share of cited appearances in AI answers for a fixed prompt set (often weighted by position). Mention rate counts any name-drop, linked or not. Citation SoV is usually the better strategy metric because linked citations drive traffic and trust; mention rate can look healthy while you still lose the cited slot. Use both, but decide content priorities from citation SoV gaps. See also AI citation share.

Is there a calculator for working out share of voice across AI answers?

Yes — compute SoV as your citations (or mentions) divided by total citations across you + named competitors on the same prompt set, then average across prompts and engines. A simple sheet works; platforms that track ChatGPT, Perplexity, Gemini and Google AI can automate the math. Kollavo's tracking views and the free LLM visibility checker help you gather the inputs. Recalculate on a weekly or biweekly cadence so one viral answer does not distort the trend.

What is AI share of voice?

AI share of voice is the percentage of all brand mentions in AI answers, for a fixed set of prompts, that go to your brand rather than to the competitors you track. You run the same prompts in ChatGPT, Perplexity, Gemini and other engines, count every brand named, and divide your mentions by the total.

We are losing share of voice in Perplexity and Claude to competitors. What can we do?

First find the prompts where competitors are named and you are not, and check which sources those answers cite. Then work on two fronts: publish clearer, more extractable answers for those exact questions on your own site, and get your brand onto the review sites, listicles and publications the engines already cite. Re-run the same prompt set weekly or biweekly to see whether your share recovers.

Related reading