- AI visibility is measured across a fixed set of buyer prompts, not single keywords.
- Core metrics: mention rate, share of voice, recommendation rate, citation rate, sentiment and accuracy.
- Run each prompt several times and across engines, because AI answers vary between runs.
- Track AI referral traffic in analytics and add a how did you hear about us field to demo forms.
- Choose tools that show raw answers and sources, cover your engines and let you export data.
Your buyers are asking AI assistants which vendors to consider. If you are not tracking what those assistants say, you are flying blind on a channel that increasingly shapes shortlists before anyone visits your site.
Here is how to measure your brand's visibility in ChatGPT, Perplexity, Google AI Overviews and other AI engines, with or without a dedicated tool.
Why AI visibility needs its own tracking
Rank trackers measure positions on a results page. AI answers have no fixed positions, change between runs and often mention brands without linking to them. A brand can rank first on Google for a category term and still be missing from the AI answer to the same question. You need metrics built for answers, not pages.
The metrics that matter
| Metric | What it tells you | How to calculate |
|---|---|---|
| Mention rate | How often you appear at all | Prompts that mention you ÷ total prompts |
| Share of voice | Your visibility against competitors | Your mentions ÷ all brand mentions in the set |
| Recommendation rate | How often you are a top pick, not just listed | Prompts where you are recommended ÷ total prompts |
| Citation rate | How often your own pages are cited as sources | Answers citing your domain ÷ total answers |
| Sentiment | How you are described | Positive, neutral or negative per mention |
| Answer accuracy | Whether facts about you are right | Correct claims ÷ claims checked |
How to build your prompt set
Start with 30 to 100 prompts that mirror how real buyers ask:
- Category prompts: best [category] tools for [company type].
- Comparison prompts: [you] vs [competitor], alternatives to [competitor].
- Problem prompts: how to solve [the problem your product solves].
- Brand prompts: what is [your brand], is [your brand] good for [use case], [your brand] pricing.
- Persona variants: the same questions phrased by a CISO, an engineer or an IT manager.
Keep the core set stable so month-over-month comparisons mean something, and test new prompts in a separate group.
How prompt tracking works
AI visibility tools send your prompts to each engine on a schedule, store the full answers and parse them for brand mentions, position in the answer, sentiment and cited sources.
Because language models give different answers to the same prompt, good tools run each prompt several times, and in different locations where relevant, then report averages. Expect some week-to-week volatility and judge monthly trends, not single snapshots.
Track AI referral traffic in analytics
Some AI engines pass referral data when users click a cited link. In GA4, create a custom channel group or exploration for referrers such as chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com and claude.ai.
Clicks from Google AI Overviews and AI Mode are reported inside Google organic traffic, so they do not appear as a separate referrer.
Finally, add a self-reported attribution field such as how did you hear about us to demo forms. Buyers increasingly answer ChatGPT or an AI assistant, which no referrer data captures.
What to look for in an AI visibility tool
- Engine coverage: ChatGPT, Perplexity, Google AI Overviews and AI Mode, Gemini, Claude and Copilot.
- Raw answers and sources visible for every prompt.
- Repeated runs per prompt and a clear sampling method.
- Competitor benchmarking and share of voice by topic.
- Sentiment and accuracy analysis.
- Exports or API access so you own your data.
- Location and language options if you sell in several markets.
Do not choose on dashboards alone. Run a trial on your own prompts and spot-check the answers against what you see in each engine.
A manual tracking method to start today
No budget for a tool yet? Start with a spreadsheet:
- List 25 core prompts.
- Once a month, run each one in ChatGPT, Perplexity and Google in a clean browser session.
- Record whether you are mentioned, your position, competitors mentioned, cited sources and any wrong claims.
- Calculate mention rate and share of voice.
- Note the sources that appear repeatedly and add them to your outreach plan.
It takes a few hours a month and gives you a real baseline before you invest in tooling. To act on what you find, read GEO vs SEO.
FAQs
Co-founder and SEO lead at OGAI. Dave builds technical SEO and content programs that turn organic search into pipeline for B2B cybersecurity and SaaS companies.