To track your AI share of voice, run a fixed set of category and recommendation questions across ChatGPT, Claude, Perplexity, and Gemini on a schedule, and record what percentage of those answers mention your brand versus each competitor. Do it monthly, never change the questions, and the trend line is your scoreboard, no paid tool required to start.
AI share of voice is the metric that replaces the rankings report as search turns into answers. This is the exact method I use before recommending any tooling, and it grows directly out of the measurement layer in the AEO + GEO playbook.
What AI Share of Voice Is
AI share of voice is the percentage of relevant AI-generated answers that mention your brand compared with your competitors, across assistants like ChatGPT, Perplexity, and Gemini. If ten buying-intent questions in your category produce answers that name a competitor eight times and you twice, your share of voice is 20%.
It is the direct descendant of share of voice in traditional media and paid search, moved to the surface where research now happens: the answer box and the chat window. The difference is that no ratings panel reports it for you, so you have to measure it yourself.
Why It Replaces the Rankings Report
Rankings are becoming a worse proxy for visibility. When an answer resolves the query in place, position one can still lose the user. Pew Research found click-through drops from 15% to 8% when a Google AI Overview appears, so a page can rank and still go unseen.
Share of voice measures what actually matters now: are you in the answer at all? It is the number that moves when your entity gets clearer and your content gets more citable, even while sessions in analytics stay flat.
Influence can grow while traffic stays flat. If you only measure sessions, you will miss the entire shift.
Building the Query Set That Matters
The whole exercise is only as good as the questions. A useful set covers the three ways buyers actually query an assistant, and it stays fixed once you start.
- Category queries — “best CRM for a small team,” “top project management tools.” These test whether you appear in the consideration set at all.
- Recommendation queries — “what should I use to send invoices,” “recommend a tool for X.” These test whether the model actively names you.
- Direct brand queries — “is [brand] any good,” “what does [brand] do.” These test accuracy, not just presence.
- Comparison queries — “[you] vs [competitor].” These show exactly where you win and lose head-to-head.
Fifteen to thirty questions is plenty. Fewer than that and one odd answer swings the whole number; many more and the monthly run becomes a chore you skip.
The Free Method, Step by Step
You can build a real share-of-voice tracker today with a spreadsheet and an hour a month.
- 1Lock the query set from the categories above, and write it down so it never drifts.
- 2Run every query on each assistant.ChatGPT, Claude, Perplexity, and Gemini. Use a clean session each time so prior chat history doesn’t skew results.
- 3Log the outcome per query, per assistant. Were you mentioned? Which competitors appeared? Was the description accurate? What was the sentiment?
- 4Note the sources. On Perplexity especially, record which pages it cited, that tells you where the answer is coming from and where to intervene.
- 5Re-run the identical set next month and compare the percentages.
How to Score It: Three Layers, Not One
Track three things, because a raw mention rate hides as much as it reveals.
- Mention rate — the headline percentage of answers that name you.
- Accuracy — whether the model describes you correctly when it does mention you. If not, you have a brand-description problem to fix.
- Sentiment — how favourably you’re characterised. Answer engines weight the written sentiment in reviews heavily, so a lukewarm mention is not the same as a strong one.
A mention that describes you wrong is not a win. Track accuracy and sentiment alongside the raw mention rate, or the headline number will mislead you.
Turning the Data Into Action
The score is only useful if it changes what you do next. Read the log for patterns, not just the top-line number.
When a competitor keeps appearing and you don’t
That is a consensus gap. Find the third-party lists and reviews the assistant is drawing on and get accurately represented in them. This is the most common reason ChatGPT recommends a competitor over you.
When you appear but the description is wrong
That is an entity problem, not a content-volume problem. Reconcile your own pages and profiles so the web tells one consistent story about you.
When nobody relevant is cited
That is an open gap. Publish the clean, answer-first content that resolves the question and you can become the default source, exactly the formats shown in examples of content that gets cited by AI.
Reading the Benchmark: What Good Looks Like
A share-of-voice number means nothing in isolation; it only means something against a baseline and a set of competitors. The first month you measure is not a grade, it is a starting line. What matters is the direction of travel and your position relative to the brands you compete with.
As a rough orientation, treat a mention rate under roughly ten percent as an entity or consensus problem, the model barely knows you exist for these queries. A rate in the middle band means you are in the consideration set but not the default answer, which is usually a clarity and coverage issue. A high rate with a strong competitor still appearing alongside you means the work shifts from visibility to differentiation. These bands are not precise thresholds; they are a way to read the number as a diagnosis rather than a score.
Segment the benchmark by query type, too. It is common to score well on direct brand questions, because your own pages answer those, while scoring poorly on category and recommendation questions, because those depend on third-party consensus you have not built. That split tells you exactly where to spend: the category queries are where recommendations, and revenue, are won or lost.
Finally, watch accuracy as its own line. A brand can climb from twenty to forty percent mention rate while its descriptions stay wrong, which looks like progress but converts badly, because the model is recommending a version of you that does not match reality. If that is happening, the fix is correcting how AI describes your brand before pushing for more mentions.
Mistakes That Ruin the Data
Monthly is the right cadence for most brands, with a deeper look each quarter; for how the monthly check fits alongside a full review, see how often to run an AI visibility audit. A few habits quietly destroy the whole exercise:
- Changing the query set. The single biggest mistake. Rewrite the questions and you lose comparability. Freeze the queries, vary only the date.
- Running in a personalised session that biases the model toward you.
- Tracking only mention rate and missing that the mentions are inaccurate or negative.
- Measuring once and treating a single snapshot as a trend. The value is entirely in the movement over time.