AI Share of Voice: How to Calculate It, and What a Good Number Looks Like

AI share of voice is your brand's share of all brand appearances, or of all picks, in AI answers to a fixed set of questions about your category. It works like classic share of voice, with AI answers in place of ads or rankings.

It comes in two versions. Presence share counts every answer that names your brand, and pick share counts only the answers that choose it. A brand named in every comparison can lead on presence and trail on picks, so report both.

To calculate AI share of voice, start with a fixed set of buyer questions, run on each AI surface you care about. Then, for every answer, record which brands appear and which are actually chosen. Two shares come out of that:

• Presence share: your appearances divided by all brand appearances. An appearance is any answer that names you in its recommendations or its lists.
• Pick share of voice: answers that pick you as their own choice, divided by all brand picks across the same answers. (You can also count recommendations, meaning any stated use case, but in our data almost every appearance carries one, so picks separate brands far better.)

Count each brand once per answer. A brand named five times in one answer is still one appearance. That rule alone removes most of the noise in homemade dashboards.

The AI share of voice benchmarks for these 21 software categories come from our published teardowns: 100 buyer questions per category, ChatGPT with web search, every answer coded by two independent AI readers against one protocol (codebook v2.0).

Two things stand out. First, few brands dominate: only Bitwarden, in password managers, held more than a quarter of the picks, and between 18 and 48 brands appeared in each category. Second, picks are more concentrated than appearances. In 20 of 21 categories the top three took a bigger share of the picks than of the appearances. AI answers name a wide field and then choose from a narrower one.

Counted per answer rather than as a share of picks, the same data reads as brand salience in AI answers: the most-picked brand in each category was ChatGPT's own pick in a median of 61 of its 100 answers.

Four mistakes distort share of voice: two in how brands are counted and two in how questions are chosen and run.

• Counting mentions instead of picks. A brand named in every comparison can lead on presence and trail on picks. See our brand mentions guide for real examples.
• Counting the brand in the question. In "alternatives to X", X is the thing being replaced. Counting it as a recommendation inflates the incumbent. We code it separately and exclude it.
• Changing the question set. Add or drop questions and the share moves for reasons that have nothing to do with your brand. Keep a fixed core set and add new questions as a separate group.
• One run per question. AI answers vary between runs. Treat single-run numbers as a snapshot and look at trends.

The share of voice number tells you where you stand. The useful work is in the breakdown: which question types you lose, which competitor takes them, and which sources those answers cite. That is the analysis behind the AI SEO services we run. For the full measurement method, including citation share, see AI visibility.

These AI share of voice questions cover the formula, what a good number is and why tools disagree.

How do you calculate share of voice in AI search?

To calculate share of voice in AI search, count the answers in which each brand is the answer's own pick across the same question set, then divide your count by the total picks for all brands. Do the same with appearances to get presence share. Report both.

What is a good AI share of voice?

A good AI share of voice is relative. In our 21 software categories, the most-picked brand held between 12 and 32 percent of all picks. Above 20 percent put a brand at the very top of its category. Compare yourself with named competitors on the same questions rather than with a universal target.

Why do different tools report different AI share of voice?

AI share of voice varies from tool to tool because each one runs different questions on different schedules and counts mentions differently. A share is only comparable when the question set, the surfaces and the counting rules stay fixed.

• High Salience research: 21 software teardowns with datasets
• High Salience: Benchmark 01 methodology

Every High Salience engagement starts with the Category Salience Brief, which shows your AI share of voice in your own category: your commercial questions run across Google and the AI surfaces, competitors side by side, and a ranked list of what to fix first.