What Is AI Share of Voice and How Do You Measure It?

The short version: AI share of voice is the percentage of relevant AI chatbot and AI search answers that mention your brand instead of your competitors, and you measure it by running a fixed set of prompts regularly and counting the mentions by hand or with a tracking tool. It's a useful directional signal, not a precise science, and anyone telling you otherwise is selling you something.

What AI share of voice means

Share of voice used to mean one thing: how much of the conversation in your industry was about your brand versus everyone else's, measured across PR mentions, ad spend, or social chatter. AI share of voice is the same idea pointed at a new place, the answers that ChatGPT, Perplexity, Google's AI Overviews, and Claude give when someone asks a question your business could answer.

If someone types "best project management tool for a 10-person agency" into ChatGPT, does it mention your product? If they ask Perplexity "who are the top AI consultants in the UK", does your name come up? That's the whole concept. It's not about rankings on a page. There's no page. It's about whether you exist in the answer at all.

This matters because a growing share of research and buying decisions never touch a search results page any more. People ask the chatbot directly and act on whatever it tells them. If you're invisible in that answer, you've lost the sale before the person even knows your website exists.

Why this isn't the same as SEO rank

With traditional SEO you had a stable, checkable thing: position one to ten, tracked daily, same results for everyone searching the same term in the same country. AI answers don't work like that. Ask the same model the same question twice in the same afternoon and you can get two different lists of brands mentioned. Ask it from a different account, a different location setting, or a slightly different model version, and the answer shifts again.

This is the bit most writers covering this topic skip past. AI share of voice tools will sell you a clean percentage, "you have 23% share of voice in your category", as if it's as solid as a Google ranking. It isn't. The underlying system is probabilistic, it changes between sessions, and the companies selling you the dashboard rarely explain that the number you're looking at is an average of samples, not a fixed fact. Treat it the way you'd treat a poll, directionally useful, wrong in the detail, right in the trend over months.

A worked example

Say you run an accounting software business aimed at UK sole traders, competing against FreeAgent, QuickBooks, and Xero. To build a basic AI share of voice measure, you'd pick 40 to 50 prompts real customers might type, things like:

• "best accounting software for self employed UK"
• "is FreeAgent worth it for a sole trader"
• "cheapest alternative to QuickBooks for freelancers"
• "accounting software that connects to my bank UK"

You run every prompt through ChatGPT, Perplexity, and Google's AI Overview once a month, on a fresh session each time so prior chat history doesn't skew the answer. You log whether your brand appears, whether it's named first, second, or buried in a list, and whether each competitor appears too.

Out of 50 prompts across three platforms, that's 150 checks a month. Say your brand shows up in 34 of them, QuickBooks shows up in 98, Xero in 71, FreeAgent in 52. Your AI share of voice is roughly 23% (34 divided by 150), against QuickBooks's 65%. That gap tells you something concrete: QuickBooks is dominant in the training data and in the content AI models are pulling from, likely because of volume and age of content, review coverage, and brand mentions across the web. Your job over the next six months is to close that gap with consistent, citable content, not to panic and buy a tool that promises to "fix" it in a week.

Step by step: how to measure it yourself

You don't need expensive software to start. Here's the manual process:

• Step 1. List 30 to 50 real prompts your buyers would type, pulled from actual sales calls, support tickets, or review sites, not guesses.
• Step 2. Pick your platforms. At minimum cover ChatGPT, Perplexity, and Google AI Overviews, since those three cover most of the AI search traffic right now.
• Step 3. Run each prompt in a logged-out or fresh session so personalisation doesn't inflate your own results.
• Step 4. Log three things per prompt, result: mentioned yes/no, position in the list, and which competitors appeared alongside you.
• Step 5. Repeat monthly, same prompts, same method, and track the trend in a simple spreadsheet rather than chasing a single month's number.
• Step 6. Cross-reference with your actual web mentions, you can use the same manual, no-paid-tool approach I've laid out for tracking social media success without paid analytics tools, because the discipline of consistent manual tracking transfers directly.

If 50 prompts a month by hand sounds like a lot, it is, roughly two to three hours of work depending on how fast you type. That's the actual cost of doing this without paying for software. Most businesses who skip this step either pay £200 to £500 a month for an AI visibility tracking tool, or they do nothing and guess. Neither is wrong, but know which one you're choosing.

What drives your AI share of voice up or down

AI models form their answers from patterns across huge amounts of text, so the things that influence your share of voice are mostly the same things that build authority anywhere else, just compounded. Volume of third-party mentions matters more than anything you write about yourself. A single glowing review on a site Google and the AI models trust is worth more than ten blog posts on your own site saying the same thing.

Structured, factual, specific content gets cited more than vague marketing copy. If your "About" page says "we're passionate about delivering excellence", no model is quoting that. If your page says "we process 40,000 invoices a month for UK sole traders and our average onboarding time is 11 minutes", that's the kind of specific, checkable claim that gets pulled into an answer.

Consistency of brand voice across every mention helps too, oddly, because models are pattern-matching, and a brand that describes itself the same way across its own site, its reviews, its press coverage, and its social profiles reads as more coherent and more "real" to the system than one that sounds like five different companies depending which page you land on. That's one reason having a fixed voice bank document for your brand isn't just a nice-to-have for your own writers, it keeps the version of you that ends up scattered across the web recognisably the same business.

The uncomfortable part nobody wants to say out loud

A high AI share of voice doesn't automatically mean more customers. I've watched the influencer world go through this exact argument for a decade, big follower counts and loud mentions that don't convert into a single sale. The same trap exists here. Being named in an AI answer is exposure, not revenue. If the AI mentions your brand but a competitor's product is the one recommended, or your brand is mentioned only in a "here are five other options" throwaway line at the bottom, that's a mention with almost no commercial value.

This is the same lesson from measuring influencer metrics that matter instead of vanity numbers, a big number on a dashboard means nothing if it doesn't connect to pipeline. Track whether people who come to your site after an AI chat session convert, not just whether you got mentioned. Most AI SOV tools can't tell you that yet, which is another reason to treat the percentage as one input, not the whole scoreboard.

Building this into your regular marketing reporting

If you already run a monthly or quarterly reporting rhythm for social and content, the simplest move is to fold AI share of voice into that same reporting cycle rather than treating it as a separate project. The same discipline I talked through in the webinar on measuring social media success applies directly here, pick a small number of metrics, track them the same way every time, and resist the urge to switch methods every month because the number didn't move the way you wanted.

If you're automating content production to keep up with this, which most businesses now are, make sure the automation doesn't flatten your brand into the same generic voice every other company is using, because generic content is exactly what gets ignored by both readers and AI models looking for something distinctive to cite. There's a reason I wrote about how to automate your marketing without losing your brand voice, the businesses winning AI share of voice right now are the ones whose content still sounds like somebody, not a template.

If building and maintaining this tracking feels like more than you want to take on inside a busy marketing team, that's a reasonable place to bring in outside help to set the system up once and hand you something repeatable, which is the kind of practical, hands-on work covered on my AI implementation coaching page.

What to do with the number once you have it

Don't chase the percentage for its own sake. Use it to answer three practical questions: are we visible at all in our category's AI answers, are we losing ground to a specific competitor across multiple prompts, and is our visibility concentrated in one platform while we're invisible in the others. Those three answers point to three different fixes, more third-party coverage, more specific competitive comparison content, or platform-specific optimisation, and none of them get solved by buying a dashboard and staring at a number that moves around because the underlying system is built to vary.

Frequently asked questions

Is AI share of voice the same as AI search ranking?

No. AI search ranking is roughly where you sit in a list, if the AI model returns one. AI share of voice is the broader measure of how often you're mentioned at all, across however many prompts and platforms you're tracking, which makes it closer to a visibility percentage than a position.

How often should I measure AI share of voice?

Monthly is enough for most businesses. Because the same prompt can return different results session to session, checking weekly mostly adds noise rather than insight. Monthly, using the exact same prompt list each time, gives you a trend you can trust.

Do I need a paid tool to track AI share of voice?

No, you can do it manually with a spreadsheet, 30 to 50 prompts, and a couple of hours a month running them through ChatGPT, Perplexity, and Google's AI Overviews. Paid tools automate the checking and add historical trend graphs, but the underlying method is the same one you can run yourself for free.

Why does my competitor dominate AI answers when their product is worse?

AI models cite based on volume and consistency of mentions across the web, not product quality. A competitor with more reviews, more press coverage, and older, more cited content will usually dominate AI answers even if your product performs better, which is exactly why closing the gap takes sustained third-party coverage rather than a single rewritten web page.