How do you calculate AI share of voice?

Run a fixed set of prompts several times on each engine. Count the answers that mention each brand. Divide your count by the total mentions of all tracked brands. Citation share works the same way using cited domains instead of named brands.

What is a good AI share of voice?

There is no universal benchmark. It depends on how many competitors the category has and how many brands engines tend to list. Compare yourself to named competitors on the same prompts and watch the trend over time.

Is citation share the same as share of voice?

They are related but different. Share of voice usually counts brand mentions in the answer text. Citation share counts which domains are linked as sources. A brand can lead one and trail the other.

Should I combine engines into one share of voice number?

It is better not to. ChatGPT, Gemini, Perplexity, Claude and Grok pick sources differently and have different audiences. A blended figure can hide a strong result on one engine and a weak one on another.

AI search glossary

AI share of voice: your slice of the brand mentions in AI answers

AI share of voice is the portion of brand mentions or citations in AI-generated answers that go to one brand, compared with its competitors, across a defined set of prompts, engines and time period.

Also called: share of voice AI, citation share, LLM share of voice, share of model

Why it matters for AI search

In a category answer, AI engines usually name only a few brands, so share of voice in AI answers is closer to a share of the shortlist than a share of attention. Because answers change between runs, the number only means something when it comes from many prompts and repeated runs, reported per engine with a range.

Where the term comes from

Share of voice started in advertising. Wikipedia describes it as the portion of ad inventory or spend one advertiser holds in a market over a period, worked out as your spend divided by total category spend. A brand spending $5 million in a $100 million category has a 5% share of voice.

The AI version keeps the idea but changes the unit. Instead of ad spend, it counts how often each brand is named, or each domain is cited, in AI answers to the same set of questions.

How to calculate it

Pick 20 to 50 prompts real buyers would ask, such as "best accounting software for freelancers".
Choose the competitors to track.
Run each prompt several times on each engine, because the answers vary.
For each engine, count mentions per brand and divide by total mentions of all tracked brands.

Example: across 100 answers on one engine, tracked brands are mentioned 250 times and yours appears 40 times. Your share of voice on that engine is 16%. Your mention rate, the share of answers that name you at all, is 40%. Both numbers are useful, and they answer different questions.

Why a single number misleads

AI answers are generated fresh each time and may search different sources. Run the same prompt twice and the list of brands can change. A share of voice from one run per prompt is mostly noise. Report a range, keep the prompt set stable, and do not blend engines.

The prompt set also shapes the result. Share of voice on "best project management software" says little about a niche tool that wins "project management for architects". Pick prompts that match the buyers you actually serve, and weight the ones closest to a purchase decision. Changing the prompt list between periods breaks the comparison, so treat it like a fixed panel. GetPicked reports a pick rate per engine with a likely range rather than a rank; the methodology page shows how.

What moves it

Share of voice in AI answers follows the sources engines read: your own pages, reviews, comparison articles, forums and press. Fixing gaps in those sources is the work. See brand mentions and AI citations for the underlying signals, AI visibility for the broader picture, and tools that track it such as Peec AI and Profound.

Questions

How do you calculate AI share of voice?: Run a fixed set of prompts several times on each engine. Count the answers that mention each brand. Divide your count by the total mentions of all tracked brands. Citation share works the same way using cited domains instead of named brands.
What is a good AI share of voice?: There is no universal benchmark. It depends on how many competitors the category has and how many brands engines tend to list. Compare yourself to named competitors on the same prompts and watch the trend over time.
Is citation share the same as share of voice?: They are related but different. Share of voice usually counts brand mentions in the answer text. Citation share counts which domains are linked as sources. A brand can lead one and trail the other.
Should I combine engines into one share of voice number?: It is better not to. ChatGPT, Gemini, Perplexity, Claude and Grok pick sources differently and have different audiences. A blended figure can hide a strong result on one engine and a weak one on another.