How-toGenerative engine optimization
How to Measure GEO Results: Metrics, Tools, Reporting
by the AEO GEO Labs team6 min read
The core GEO metrics are mention rate, citation rate and AI share of voice, measured on a fixed set of buyer prompts across the engines your buyers use. Add first-party data from Google Search Console's Generative AI report and Bing's AI Performance report, plus AI referral sessions in analytics. Together they show whether AI engines see you, cite you, prefer you, and send you business.
tl;dr
- 1Build one fixed prompt set and never change it mid-quarter.
- 2Run each prompt more than once; AI answers vary.
- 3Use Google and Bing's free reports for the data only the engines have.
- 4Tie it to business outcomes with AI referral sessions and pipeline.
This guide covers what to measure, how to collect it, and how to report it. It builds on our complete guide to generative engine optimization.
The GEO metrics that matter
Six GEO metrics cover visibility, preference and outcomes. The first three come from prompt tracking; the rest come from first-party reports and analytics.
| Metric | What it measures | How to calculate or find it |
|---|---|---|
| Mention rate | How often AI answers name your brand | Answers mentioning you ÷ total answers for your prompt set |
| Citation rate | How often AI answers link to your pages | Answers citing your domain ÷ total answers |
| AI share of voice | Your mentions relative to competitors | Your mentions ÷ mentions of you plus named competitors |
| Description accuracy | Whether engines describe you correctly | Share of answers with correct category, pricing and key facts |
| AI impressions and citations | How often engines showed or cited your pages | Google Search Console Generative AI report; Bing AI Performance |
| AI referral sessions | Visits from AI engines | Analytics sessions from chatgpt.com, perplexity.ai and similar sources |
Two more are worth watching but are harder to attribute: branded search volume, which can rise when AI engines recommend you without a click, and self-reported attribution, such as a "How did you hear about us?" field that offers AI assistants as an option.
For deeper definitions, see AI visibility metrics: what to measure and what to ignore.
Step 1: Build a fixed prompt set
A fixed prompt set is the foundation of GEO measurement, because every rate above is calculated against it. Without a stable set, month-on-month changes are noise.
- Write 30 to 150 prompts in the words buyers actually use, drawn from sales calls, support tickets and search questions.
- Cover every stage: problem, category, comparison and decision questions.
- Keep branded prompts separate. "What is [your brand]?" measures accuracy; "best X for Y" measures visibility. Report them apart.
- Name three to five competitors to benchmark against.
- Choose engines and markets. Most B2B teams start with ChatGPT, Google AI Overviews, Perplexity, Gemini and Microsoft Copilot in one country.
- Freeze the set for the quarter. Add new prompts as a separate group so trends stay comparable.
Step 2: Collect the data
You collect GEO data from three sources: prompt runs, first-party engine reports and your own analytics.
Prompt runs
Prompt runs are the only way to see ChatGPT, Perplexity and Claude answers. Run each prompt at least two or three times per period, because the same prompt can return different answers. Record whether you were mentioned, whether you were cited, the cited URL, competitors named, and how you were described. A spreadsheet works for up to about 50 prompts a month; beyond that a tracking tool saves real time. Our guide to tracking AI brand mentions covers both routes.
Google Search Console
Google's data comes from the Generative AI performance report. As of October 2026, Google's help page for the report says it shows impressions from AI Overviews and AI Mode, grouped by page, country, device and date, and it was rolled out to all sites on 31 August 2026. It reports impressions, not clicks, and the same data also sits inside the overall Performance report.
Bing Webmaster Tools
Microsoft's data comes from the AI Performance report, which Bing introduced in public preview in February 2026. It shows total citations, average cited pages, grounding queries (the phrases the AI used to retrieve your content) and page-level citation counts across Microsoft Copilot and AI summaries in Bing.
Analytics
Your analytics shows whether AI visibility turns into visits. Create a segment or custom channel for AI referrers. A starting regex for the session source:
chatgpt\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai
Clicks from ChatGPT commonly arrive tagged with utm_source=chatgpt.com. Expect AI referrals to undercount real influence: many people read an AI answer, then search your brand directly.
Step 3: Calculate share of voice
AI share of voice is your share of brand mentions among a defined competitor set, across your prompt set. It is the best single GEO metric for executives, because it is relative and survives changes in how often engines name any brand at all.
The calculation:
AI share of voice = your mentions ÷ (your mentions + all named competitors' mentions)
Example: across 100 prompt runs, you are mentioned 18 times and four competitors are mentioned 22, 30, 15 and 15 times. Your share of voice is 18 ÷ 100 = 18%. Calculate it per engine as well as overall, because the gaps between engines show where to focus. The full method, including weighting by position, is in how to calculate AI search share of voice.
Step 4: Report monthly
A monthly GEO report should fit on one page and answer four questions: are we more visible, are we preferred over competitors, are we described correctly, and is it driving business?
| Section | Contents |
|---|---|
| Visibility | Mention rate and citation rate, overall and per engine, vs last month and baseline |
| Competition | Share of voice vs named competitors, per engine |
| Accuracy | Errors found in AI descriptions of you, and fixes shipped |
| First-party | Google AI impressions by top page; Bing citations and grounding queries |
| Outcomes | AI referral sessions, conversions, branded search trend |
| Work shipped | Technical, content and off-site changes made this month |
Report rates with their sample size. "Mentioned in 24 of 120 runs" is honest; "visibility up 20%" off a handful of prompts is not.
Warning: don't read meaning into a single month's swing. Engines update models and retrieval often. Look for trends across at least two or three months before changing strategy.
Before you start tracking, our free AI visibility grader gives a quick read on whether your pages are ready to be cited.
Frequently asked questions
What are the most important GEO metrics?
The most important GEO metrics are mention rate, citation rate and AI share of voice, all measured on a fixed set of buyer prompts across the engines your buyers use. Add description accuracy, first-party data from Google Search Console and Bing Webmaster Tools, and AI referral sessions from analytics to connect visibility with business results.
How do I track GEO results for free?
Write a fixed list of 30 to 50 buyer prompts, run them monthly in ChatGPT, Google AI Overviews and Perplexity, and log mentions, citations and competitors in a spreadsheet. As of October 2026, add Google Search Console's free Generative AI performance report and Bing Webmaster Tools' AI Performance report, plus an AI referral segment in your analytics.
Does Google Search Console show AI Overviews data?
Yes. As of October 2026, Search Console has a Generative AI performance report showing impressions for AI Overviews and AI Mode, by page, country, device and date. Google rolled it out to all sites on 31 August 2026. It shows impressions rather than clicks, and the same data also remains inside the overall Performance report.
How do I measure GEO results in ChatGPT?
ChatGPT offers no publisher dashboard, so you measure it two ways. Run a fixed prompt set in ChatGPT regularly, several times per prompt, and log mentions and citations. Then track referral sessions in analytics, where ChatGPT clicks usually appear with chatgpt.com as the source, often tagged with utm_source=chatgpt.com.
How long before GEO metrics improve?
Technical fixes, such as unblocking AI crawlers, can lift citation rates within weeks of recrawl. Content and reputation work usually takes a few months to show consistently. Because AI answers vary between runs, judge progress over two to three months of tracking against your baseline rather than on any single week.
If you want GEO tracking and reporting handled for you, AEO GEO Labs runs GEO, AEO and SEO programs for B2B and SaaS teams, starting with a free AI visibility report. See our GEO and AEO services.