Quick answer: AI share of voice is the percentage of AI-generated answers in your category that name your brand, compared with your competitors. To measure it, build a fixed set of the questions your buyers actually ask, run them on a regular schedule across ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews, and record who gets named, how each brand is described and which sources the AI cites.
In our recent posts we looked at how PR shapes what AI says about your company and how tech brands earn citations in AI search. This post answers the question our clients ask next: how do we know it's working?
Why does AI share of voice matter now?
In 2024, Gartner predicted that traditional search engine volume would drop 25% by 2026 as people turn to AI chatbots and virtual agents. Whatever the final number turns out to be, the direction is clear.
For buyers in technology, space, healthcare and telecom, the shortlist now often forms inside an AI answer before anyone visits a vendor's website. Your web analytics can't see that moment. Google Search Console doesn't break out how often you appeared inside an AI Overview, and ChatGPT doesn't send you a report. If you aren't measuring AI answers directly, you're guessing.
What is AI share of voice?
AI share of voice is the AI-era version of a metric PR teams have used for decades. Instead of counting press mentions or ad impressions, it counts how often AI assistants name your brand when someone asks a question about your category.
Most teams track two numbers:
Presence rate: the share of your tracked questions where your brand appears at all. If you appear in 18 of 50 answers, your presence rate is 36%.
Mention share: your brand mentions divided by all brand mentions across your competitive set, multiplied by 100. This is the true share of voice figure, because it shows how much of the conversation you own compared with rivals.
There is no single industry-standard formula yet. That's fine, as long as you use the same method every time so the trend line is honest.
How do you measure AI share of voice? A six-step method
Step 1: Define your competitive set
Pick five to eight competitors your buyers actually compare you with. Include the category leader even if you think you aren't competing head to head. AI assistants will mention them either way.
Step 2: Build a prompt set from real buyer questions
Write 30 to 50 prompts in the words your customers use. Pull them from sales calls, RFPs, support tickets and analyst inquiries. Cover four kinds of questions:
Category questions: "What are the best platforms for [your category]?"
Problem questions: "How do I solve [the problem your product fixes]?"
Comparison questions: "[Competitor] vs. alternatives for [use case]"
Brand questions: "What is [your company] known for?" and "Is [your company] a good choice for [use case]?"
Brand questions tell you whether AI describes you accurately. Category and problem questions tell you whether you get recommended at all. The second group matters more.
Step 3: Run every prompt across the major AI engines
Test ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews or AI Mode. Run each prompt more than once, because the same question can return different brands on different days. Use a fresh session with no chat history so past conversations don't skew the answer.
Step 4: Score each answer
For every answer, record:
Whether your brand is named
Where it appears, since the first brand named tends to get the most attention
Whether the description is accurate and current
How you are framed: recommended, mentioned neutrally, or cautioned against
Which sources the AI cited or linked
Step 5: Calculate and benchmark
Work out presence rate and mention share for your brand and every competitor, overall and by engine. Your position can differ sharply between ChatGPT and Perplexity, and that gap tells you where to focus.
Step 6: Track monthly and connect it to business results
Re-run the same prompt set every month, and weekly around launches or news moments. Then look for related signals: branded search volume, direct website traffic, AI referral traffic in Google Analytics 4, and new leads who say "I found you through ChatGPT."
What actually moves your AI share of voice?
The single most useful step is to study the citations. For every prompt where a competitor appears and you don't, read the answer and check the sources it pulled from. That list becomes your PR target list.
In our work with technology and space clients, the same levers come up again and again:
Earned media in trusted outlets. Trade and business publications are heavily used sources for AI answers. Coverage there does double duty.
Third-party lists and analyst mentions. "Top vendors" roundups, analyst reports and award lists shape category answers.
Expert commentary and bylines. Quotes and bylined articles from your executives give AI something specific and attributable to repeat.
Clear owned content. Plain-language product pages, FAQ sections and an about page that states what you do, for whom, in one sentence.
Consistent facts everywhere. Your company name, description, leadership and key numbers should match across your website, LinkedIn, Crunchbase and press materials.
Structured data. Schema markup such as Organization, Article and FAQPage from Schema.org helps machines read your pages correctly.
This is why we treat GEO, or generative engine optimization, as a PR discipline first. AI assistants mostly repeat what credible third parties say about you. You can't fake that with on-page tricks. For the search side of the story, see our post on how AI Overviews affect SEO.
Common AI share of voice mistakes to avoid
Measuring from a screenshot. One answer on one day is an anecdote, not a metric.
Using too few prompts. Five questions can swing wildly. Aim for at least 30, and more if your category is broad.
Testing only branded questions. Your brand will usually show up when someone types your name. The real test is whether you appear when they don't.
Ignoring accuracy. Being mentioned with an outdated product name, an old executive or the wrong headquarters can hurt more than being left out.
Treating it as an SEO-only project. Most of the inputs AI relies on come from earned media, analysts and partners, which is PR territory.
Frequently asked questions
What is a good AI share of voice?
It depends on your category and how many credible competitors exist. Your number compared with your top three competitors, and whether it is rising, matters more than any absolute benchmark.
How often should we measure AI share of voice?
Monthly is a good baseline. Measure more often around product launches, funding news, crises or major industry events.
Does Google Search Console show our AI Overview visibility?
Not separately. Search Console blends AI Overview data with standard search results, so it can't show how often your brand was named inside an AI answer. You need prompt-based tracking for that.
Can PR improve how often AI recommends our company?
Yes. AI assistants build answers from sources they trust, especially news coverage, analyst content and expert commentary. A steady earned media program gives those systems more accurate, more recent material to cite.
How long does it take to see a change?
Expect movement over months, not days. Assistants that search the web live can reflect new coverage quickly. Answers drawn from a model's training data change more slowly.
What is the difference between GEO and SEO?
SEO helps your pages rank in a list of search results. GEO helps your brand get named, described accurately and cited inside AI-generated answers. The two overlap, but GEO leans much more on third-party credibility.
How GRC helps brands win in AI search
Global Results Communications has spent more than 20 years helping technology, space, healthcare and telecom companies earn coverage that builds trust. Our GEO programs start with an AI visibility audit: we build your prompt set, measure your share of voice against competitors across the major AI engines, and map the sources behind every answer. Then we put PR to work on the gaps.
Want to know how AI describes your company today? Book a 30-minute call with our team, explore our capabilities and results, or contact us.