What is Share of voice (AI)?

Share of voice (AI) is your brand's cut of all the mentions across AI answers, compared to your competitors, for the same set of questions. The quick formula: your mentions divided by everyone's mentions, times 100. If ChatGPT, Perplexity, and Gemini name brands in your category 200 times across a batch of prompts, and you get named 50 of those times, your AI share of voice is 25%.

It's a relative number, not a solo score. You can't have a high share of voice on your own. Somebody else has to have a smaller one.

Where the idea comes from

Share of voice is an old marketing term. It used to mean your slice of total ad spend or total impressions in a category. If everyone in your space spent a million bucks on ads and you spent two hundred grand, you had 20% share of voice. Simple.

The AI version keeps the same shape but swaps the ingredient. Instead of counting ad dollars or impressions, you count how often AI engines mention, cite, or recommend you versus the other names that come up. Money doesn't buy you in here. You can't pay ChatGPT to put your brand in its answer. You earn the spot with credible content, citations, and a reputation the model has picked up on.

How people actually calculate it

There's no single agreed-on formula yet. As of 2026, HubSpot, Semrush, and Profound all measure it a bit differently, which is annoying but true. The common approaches:

• Mention-based. Your brand mentions divided by total brand mentions across all your tracked prompts. The most common method.
• Citation-based. Your citations (actual links to your pages) divided by total citations. Stricter, because a name-drop doesn't count, only a real source link does.
• Entity-based. How many times you show up as a recommended option divided by the total options the AI lists.

Most teams pick one, stick with it, and watch the trend instead of obsessing over the exact percentage.

A quick example

Say you sell project management software. You pick 100 questions a buyer might ask, like "best project tool for remote teams" or "Asana alternatives." You run them through the major AI engines and tally up who gets named.

If the answers mention software brands 400 times total, and your brand shows up 60 times, your AI share of voice is 15%. A rival who shows up 120 times is sitting at 30%, double you. Now you know the gap, and you know who to chase.

Why it's worth tracking

A raw mention count tells you a little. Share of voice tells you where you stand. Going from 40 mentions to 50 sounds like progress, but if your competitor jumped from 80 to 200 in the same window, you actually lost ground. The relative view catches that.

It's also kind of zero-sum. AI engines tend to lean on a small group of trusted sources, so when one brand's share goes up, somebody else's usually drops. The pie isn't growing fast. You're fighting for slices.

For a rough sense of scale: the average brand mention rate across AI answers sits around 17% in 2026. Under 15% share usually means you've got a citation gap. The 25 to 40% range is competitive in most categories. Above 40% means you're a default answer.

One big catch

Share of voice only counts how often, not how nicely. You can own 50% of the mentions in your category and still be losing if the AI keeps calling you "overpriced" or "hard to set up." Frequency without good sentiment is a trap. So most teams track share of voice next to sentiment, not on its own.

How to improve it

You raise your AI share of voice with the same work behind AI visibility in general: answer real questions directly, back claims with stats and sources, keep your content clean and structured, and earn mentions on sites the models already trust. That's the GEO playbook. Share of voice is just the scoreboard that tells you if it's working, and whether it's working better than it's working for the other guys.

FAQ