---
title: "AI Share of Voice: The New Rank Tracking | Kitbase Blog"
description: "AI share of voice measures how often AI engines mention your brand versus competitors across sampled answers. Learn why it replaces rank tracking for AI answers and how to compute it."
canonical: https://kitbase.dev/blog/ai-share-of-voice/
---

**AI share of voice is the percentage of AI-generated answers that mention your brand, measured against the total mentions of every competitor in the same set of prompts.** If you and three rivals are named across a sample of AI answers and your brand accounts for 30% of those mentions, your AI share of voice is 30%. It's the metric that replaces keyword rank tracking in a world where AI engines answer questions directly instead of returning a list of links, because "what position do I rank" stops making sense when the output is a paragraph naming three brands, not ten ordered results.

Rank tracking assumed a stable, ordered results page you could occupy a slot on. AI answers have neither stability nor slots. Share of voice is how you keep score anyway. Here's what it measures, why it's the right metric, and how to compute it correctly.

## Why rank tracking breaks for AI answers:

Everything that made rank tracking work is missing in AI answers:

- **There's no fixed position.** An AI answer names a handful of brands, sometimes in a ranked list, sometimes not, sometimes just one. "Position 3 of 10" has no equivalent.
- **The answer changes every time.** AI engines are non-deterministic, ask the same prompt twice and the brand list can differ. as we explain in [why AI answers change every time you ask](/blog/ai-answers-non-determinism/).
- **There's no search volume.** No tool reports how many people ask ChatGPT *"best analytics for startups."* You can't weight prompts by volume the way you weighted keywords.
- **Being present isn't the whole story.** Appearing in 40% of answers sounds good, until you learn a competitor appears in 90%. Your own rate, in isolation, is missing its denominator.

That last point is the crux. Presence rate is the AI-era equivalent of a ranking, and it's essential (we cover the full metric set in [the metrics of AI visibility](/blog/ai-visibility-metrics/)). But presence rate alone doesn't tell you whether you're winning the category. **Share of voice adds the competitive denominator**: your presence relative to everyone else's.

## What AI share of voice actually measures:

Share of voice normalizes your brand's presence against the total across all tracked brands for the same prompts and time window. Concretely:

> **AI share of voice** = your brand's mentions ÷ total mentions of all tracked brands (yours + competitors), across a sample of AI answers.

A worked example. You run five prompts across four AI engines, daily, for a week here's a sample of many answers. You tally how many answers name each brand:

| Brand | Answers mentioning it | Share of voice |
|---|---|---|
| You | 60 | 30% |
| Competitor A | 80 | 40% |
| Competitor B | 40 | 20% |
| Competitor C | 20 | 10% |
| **Total** | **200** | **100%** |

Your presence looks healthy in isolation, but share of voice reveals you're second, and Competitor A owns the category's mindshare. That's a different but yet useful, conclusion than any single answer or your raw presence rate would give you. Track it over time and you get the thing rank tracking used to give you: a competitive scoreboard with a trend line.

**From sampled answers to a share-of-voice scoreboard**

```mermaid
flowchart LR
  A["Prompts × engines<br/>run repeatedly"] --> B["Extract every<br/>brand named"]
  B --> C["Tally mentions<br/>per brand"]
  C --> D["Normalize:<br/>each brand ÷ total"]
  D --> E["Share of voice<br/>+ rank over time"]
```

## How to compute it correctly:

The math is a division. The rigor is in how you sample, and this is where most manual attempts go wrong.

1. **Pick prompts that represent real buyer questions.** Category head terms, "best X for Y", "X vs Y", "alternatives to X". These are the prompts where brands actually get named. Finding them is its own discipline, see [how to find the prompts your buyers ask AI](/blog/find-prompts-buyers-ask-ai/).
2. **Sample repeatedly, not once.** Because answers are non-deterministic. Run each prompt many times (daily is a sensible cadence) so your mention counts reflect a distribution, not a coin flip.
3. **Cover multiple engines.** ChatGPT, Perplexity, Gemini, and Claude retrieve differently, so your share of voice differs per engine. Compute it per engine *and* combined for example being dominant in Perplexity but invisible in ChatGPT is a strategy signal, not a rounding error. (Each engine has its own citation logic: see [how to get mentioned by ChatGPT](/blog/how-to-get-your-brand-mentioned-by-chatgpt/) and [how Perplexity chooses its citations](/blog/how-perplexity-chooses-citations/).)
4. **Extract every brand named, including ones you don't track.** If your denominator only counts brands you thought to add, you'll overstate your share. A complete tally reads *every* company each answer names which is also how you discover competitors you didn't know were winning.
5. **Track the trend.** A single week's share of voice is a snapshot. The value is watching it move after you publish content, earn a review-site placement, or a competitor does.

The reason this is rarely done by hand: it's N prompts × 4 engines × many runs × parsing every answer for every brand, then normalizing which will be done every day. That's a pipeline, not a spreadsheet.

## How Kitbase measures share of voice:

[Kitbase AI Visibility](https://docs.kitbase.dev/ai-visibility) automates the whole computation. It queries Perplexity, Gemini, Claude, and ChatGPT daily through their official APIs with the prompts you define, and an extraction model parses each answer into every company it names then matches those names to your tracked brands. From that it computes:

- **Share of voice** that normalizes each brand's presence against the total across all tracked brands per run, with a rank over time plus a competitive share, not just your own rate, viewable per engine or combined.
- **A competitor leaderboard** ranking every brand you track, with each one's mention rate, citation rate, and share of voice side by side.
- **Competitor trends over time** One line per brand, so you watch your share move against rivals run over run, not just at a single point.

Two features make this genuinely low-effort to keep accurate:

**Suggested competitors.** You don't have to know your whole competitive set up front. Because Kitbase extracts *every* brand each answer names, it surfaces companies the AI mentioned that you aren't tracking yet, ranked by how many runs named each. One click adds a suggested competitor to your tracked set, and it lands on the leaderboard from the next run. This is what keeps your denominator honest: the competitors actually winning AI mindshare get pulled into the comparison instead of being missed. (More on the discovery angle in [brand vs brand AI prompts](/blog/brand-vs-brand-ai-prompts/).)

**Retroactive re-matching.** When you add that suggested competitor, rename a brand or add an alias, Kitbase re-matches your entire stored answer history against the new brand list automatically. The newly added competitor appears with its *full* historical share of voice, not a line starting at today, and it costs nothing extra: no re-running prompts, no additional AI spend. Your scoreboard is complete the moment you add a brand.

Share of voice sits alongside the rest of the AI visibility picture and presence rate, per-engine breakdown, and the [cited-domain map](/blog/which-domains-do-ai-engines-cite/) showing *where* the answers get their information. Together they turn "are we winning in AI answers" from a vibe into a tracked, competitive metric, the job rank tracking used to do for search. For the strategy that moves the number, start with the pillar guide, [What Is Generative Engine Optimization](/blog/what-is-generative-engine-optimization/).

## FAQ

**What is AI share of voice?**
It's the share of AI-generated answers that mention your brand, measured against the total mentions of all competitors across the same prompts and time window. If your brand accounts for 30% of all brand mentions in the sample, your AI share of voice is 30%.

**How is share of voice different from presence rate?**
Presence rate is the share of answers that mention *you*, your standalone visibility. Share of voice divides your presence by the total across all tracked brands, adding the competitive denominator. Presence rate tells you if you show up; share of voice tells you if you're winning.

**Why does share of voice replace rank tracking for AI?**
Because AI answers have no fixed positions, change between runs, and come with no search-volume data, the assumptions rank tracking relied on. Share of voice keeps score in a way that fits how AI answers actually work: sampled, competitive, and trended over time.

**How do I choose which competitors to include?**
Start with the rivals you know, but let the data expand the set. Because a correct tally extracts every brand each answer names, tools like Kitbase surface competitors the AI named that you weren't tracking so you can add them so your denominator reflects the real competitive field, not just your assumptions.

**How often should I measure it?**
Frequently enough to average out non-determinism, daily sampling is a good default. A single measurement is a snapshot; the value is the trend after you (or a competitor) change something.

---

*Want your AI share of voice tracked against your competitors automatically? [Start your free trial](https://app.kitbase.dev/signup/) — 7 days, no credit card required — and see where you rank across every AI engine.*