AI share of voice
AI share of voice is how often AI answers name your brand compared with your competitors, across a set of buyer prompts. If you track 100 prompts about expense software and you're named in 30 answers while your main rival is named in 60, your share is half of theirs.
What is AI share of voice?
AI share of voice is your brand's slice of all the brand mentions in AI answers about your category. You pick a set of buyer prompts, run them in tools like ChatGPT, Perplexity, Gemini and Google AI Mode, and count how often each brand gets named. Your share is your mentions divided by everyone's.
Let's say you sell expense software and you track 100 prompts, like "best expense tool for a remote team" and "expense software that syncs with NetSuite". Across those answers, your brand is named 30 times, your main rival 60 times and two smaller tools 30 times between them. That's 120 mentions in total, so your share of voice is 25% and your rival's is 50%.
The idea comes from ads and PR, where share of voice meant your slice of all the ads or press coverage in a category. The AI version asks the same question about the answers buyers read before they ever visit a vendor's site.
Why AI share of voice matters for B2B SaaS
A single AI answer is a shortlist. When a buyer asks for the best tool and gets four names, the brands that show up most across all those answers are the ones buyers hear about first. Share of voice tells you who's winning that, prompt by prompt.
It also tells you more than AI visibility alone. Being named in 30% of answers sounds good until you find out your rival is in 70%. Here's why I'd report it every month:
- It shows the gap in one number: "We're at 25% and they're at 50%" is something your CEO understands in five seconds. It turns a vague worry about AI into a target.
- It works when the whole category moves: AI tools sometimes name fewer brands per answer, or more. Your raw mention count can drop while your share holds, and share is the fairer read.
- It points at the competitors to study: The brand that gains share fastest is doing something you're not. Look at which pages and third-party sources get cited in their answers.
How to track AI share of voice
The number is only useful if you measure it the same way every month. Here's my process:
- Build a fixed prompt set: Write 30 to 50 prompts from sales calls and demos, grouped by use case and stage. Keep them the same month to month so the trend means something, and add new prompts as a separate group.
- Pick the competitors up front: List the four to six brands you meet most in deals. Count mentions for each of them, and lump everyone else into "other".
- Run each prompt more than once: SparkToro found there's less than a 1 in 100 chance that ChatGPT or Google's AI gives the same list of brands in two answers to the same prompt. Running each prompt three to five times gives you a steadier average.
- Split it by tool and topic: You might hold 40% in Perplexity and 10% in ChatGPT, or lead on "invoicing" prompts and trail on "expenses". The split tells you where to work.
- Log the sources too: Next to each answer, note which pages were cited. When a rival's share jumps, the sources usually show you why.
A prompt tracking tool runs all this on a schedule. Our list of AI rank tracking tools compares the main options.
An AI share of voice example
Here's a quarter of tracking for a hypothetical expense software company. The team runs 40 prompts, three times each, in four AI tools every month:
| Brand | July | August | September |
|---|---|---|---|
| Rival A | 48% | 46% | 44% |
| The team's brand | 18% | 22% | 27% |
| Rival B | 20% | 19% | 17% |
| Others | 14% | 13% | 12% |
The team's share grew from 18% to 27% over the quarter. In the same months, they published three comparison pages and got added to two "best expense software" roundups that Perplexity and AI Mode cite often. Their source log shows those roundups in most of the new answers that name them.
That fits what AirOps found: on buying searches like "best CRM for startups", brands are 6.5x more likely to show up through other sites than through their own. So most share of voice gains come from getting onto the pages AI already trusts.
How to measure AI share of voice well
Share of voice is easy to calculate and easy to get wrong. Here's what I'd watch to keep the number honest:
- Share by prompt group: Report share for each use case and stage, along with the total. A total that hides a 0% on your most valuable use case isn't helping you.
- Position in the answer: Note whether you're named first, in the middle or last. Being first in 20% of answers can matter more than being fifth in 40%.
- Cited share: Count how many of the cited sources are your pages. Mentions with your own page as the source are the strongest kind.
- The trend over a quarter: Month-to-month swings are normal. Look for a direction across three months or more before you change the plan.
- Pipeline next to it: Put share of voice in the same report as branded search, demo requests and "How did you hear about us?" answers. Our guide to AI search attribution shows how to connect them.
Common AI share of voice mistakes
- Tracking prompts you'll always win: "What is [your brand]?" names you every time and inflates your share. Track the prompts buyers type before they know you.
- Changing the prompt set every month: New prompts make the trend meaningless. Add them as a new group and keep the core set fixed.
- Running each prompt once: One run is one roll of the dice. Average several runs, or a lucky week will look like a breakthrough.
- Ignoring how you're described: A mention that says your tool is "best for very small teams" when you sell to mid-market still counts toward share. Read the answers and fix the source of wrong descriptions.
- Picking the wrong competitors: If you only compare yourself with tools you always beat, your share looks great and means nothing. Use the names from your lost deals.
Here's a first step you can take today. Write down the ten prompts your buyers type most often, run them in ChatGPT and count the brands in each answer. What's your share?
Read more
FAQ
How do you calculate AI share of voice?
Run a fixed set of buyer prompts, count how many times each brand in your category is named, then divide your count by the total for all brands. Multiply by 100 to get a percentage.
What's the difference between AI visibility and AI share of voice?
AI visibility is the share of answers that name you, on its own. Share of voice puts that number next to your competitors, so it tells you whether you're winning the answer or just in it.
How many prompts do you need to track AI share of voice?
Start with 30 to 50 prompts across your main use cases, and run each one several times. AI answers change between runs, so a bigger sample gives you a steadier number.
Which tools track AI share of voice?
Prompt tracking tools like Peec AI, Profound, SE Ranking and Arrowsterr run your prompts on a schedule and report share of voice against the competitors you pick. A spreadsheet and manual checks work fine for the first month.


