AI Strategy

How to Measure AI Search Visibility: AEO and GEO Metrics That Matter

Updated 7 min read

How to Measure AI Search Visibility: AEO and GEO Metrics That Matter

AI search visibility is measured in four layers: how often your brand is mentioned for a fixed question set, how often pages are cited, how much traffic AI products send, and whether branded search and enquiries grow. No one dashboard shows all four. Report trends, not single results.

Key takeaways

  • Measure presence (mentioned or not), position, accuracy, citation and referral traffic, separately.
  • Keep a fixed question set and re-run it monthly under the same conditions.
  • Referral traffic from AI products is real but usually small today. It is the least complete signal.
  • Branded search growth and "how did you hear about us" answers often capture influence that clicks miss.
  • Be honest with yourself and clients about noise. A single changed answer is not a trend.

The four layers

LayerWhat it tells youHow to get it
1. MentionsAre you named in answers to category questions?Monthly question test across AI tools
2. CitationsAre your pages used as sources?Same test, record cited URLs
3. Referral trafficDo people click through from AI products?Analytics referrals and UTM parameters
4. Business impactAre leads and brand searches growing?Search Console, CRM, intake form questions

Layer 1 and 2: The question test

This is the backbone. Build a list of 20 to 50 questions across discovery, comparison, price and brand questions. Follow the method in our AI visibility audit guide. For each question and each tool, record:

  • Mentioned: yes or no.
  • Position among the businesses named.
  • Accuracy of what was said.
  • Competitors named.
  • URLs cited as sources, and whether any are yours.

From that, calculate simple rates each month:

  • Mention rate = questions where you are named / total questions asked.
  • Citation rate = questions where one of your URLs is cited / total.
  • Accuracy rate = answers about you that are fully correct / answers that mention you.

Run important questions twice and count the better-supported result, or average them. Tools that support web search in an API make it possible to automate this. Treat automated results as a complement to manual checks, because consumer apps can differ from API output.

Layer 3: Referral traffic from AI products

In Google Analytics 4, look at Acquisition, then Traffic acquisition, and filter by source. Common patterns you may see (these change, so check your own data):

  • Source or referrer containing chatgpt.com or chat.openai.com. ChatGPT often adds a utm_source=chatgpt.com parameter to links.
  • perplexity.ai.
  • gemini.google.com or copilot.microsoft.com.
  • Claude referrals appear under claude.ai.

Create a custom channel group called "AI assistants" that includes these sources, so you can trend it. Two caveats: many AI answers produce no click at all, and some clicks arrive with no referrer, so they appear as direct traffic. The numbers undercount.

Google AI features

Google includes clicks and impressions from AI features in the overall Search Console Performance report and does not currently break them out on their own. What you can do:

  • Track impressions and clicks for your question-style queries over time.
  • Spot-check live results for your top 20 queries monthly. Note whether an AI Overview appears and whether you are linked.
  • Compare click-through rate on queries where an Overview appears and where it does not.

Layer 4: Business impact

  • Branded search volume in Search Console. People who see you in an AI answer often search your name next.
  • Direct traffic trends, which can rise when people copy your name or URL from an answer.
  • Self-reported source: add "How did you hear about us?" to your contact form and include "AI assistant (ChatGPT, Gemini, etc.)" as an option. Also ask on sales calls.
  • Lead quality: compare close rates of leads from AI referrals against other channels.

A simple monthly report

  1. Date and list of tools tested.
  2. Mention rate, citation rate and accuracy rate, with last month's numbers beside them.
  3. Three notable changes (a new competitor, a lost mention, a corrected fact), with the quoted answer.
  4. AI referral sessions and conversions from analytics.
  5. Branded search impressions and clicks.
  6. Actions taken last month and planned for next month.

Cautions

  • Variance: two runs of the same question can differ. Do not celebrate or panic over one result.
  • Personalisation: logged-in accounts and location change answers.
  • Vendor dashboards: some SEO platforms sell AI visibility scores. Ask how they collect data and how often. Methods differ widely.
  • Attribution is imperfect. Combine several signals and judge the direction.

Set up the AI assistants channel in GA4

  1. In Google Analytics 4, open Admin, then Data display, then Channel groups.
  2. Copy the default channel group and add a new channel called "AI assistants".
  3. Set the condition to Source matches regex and paste a pattern such as:
    chatgpt\.com|chat\.openai\.com|perplexity|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai
  4. Place the new channel above "Referral" so it takes precedence, then save.
  5. Check the Traffic acquisition report a few days later and adjust the pattern if new sources appear.

Source names change, so review the list each quarter against your own data.

Search Console: find your question queries

In the Performance report, add a Query filter set to "Custom (regex)" and match the question words people use:

^(how|what|why|when|which|who|can|does|is|are|should)\b|cost|price|best|vs|near me

Save the filtered view and compare clicks and impressions month on month. Question-style queries are where AI answers appear most often.

A sample monthly tracker

These figures are invented, to show the layout. They are not real results.

MetricAugustSeptemberOctober
Questions tested303030
Mention rate10%17%23%
Citation rate3%7%10%
Accuracy rate60%75%86%
AI referral sessions121931
Branded search clicks140162181

Read it as a direction. A rising mention and accuracy rate while citations lag points to the need for pages that can actually be cited. Falling accuracy points to a wrong source that needs fixing. Because answers vary between runs, keep the trend, not any single month, in your decisions.

What Google's documentation says about reporting

Google states that traffic from AI Overviews and AI Mode counts toward your overall search traffic and appears in the Search Console Performance report under the Web search type. It does not offer a separate AI-only report, so any claim of exact AI Overview traffic from Search Console data alone should be treated with caution.

Related guides

Frequently asked questions

Is there one tool that measures everything?

No. You will combine a question test, analytics and Search Console. Some paid tools bundle parts of this.

What is a good mention rate?

It depends on your market. Compare against competitors on the same question set and look at your own trend.

How many questions do I need?

20 is enough to start. 50 gives a steadier picture. Keep the list stable.

Why is my AI referral traffic so low?

Many answers satisfy the user without a click, and some clicks are not labelled. Low numbers are common. They do not mean your visibility is zero.

Can Growmerz set up monitoring for us?

Yes. We build monthly monitors and reports as part of our AEO and GEO service.

Should I include screenshots of AI answers in reports?

Yes, with the date, tool and question. They are evidence of what was shown, and answers can change.

Sources and further reading

Primary documentation and research behind this guide.

We set up monthly monitors and plain-language reports so you can see the trend without building the spreadsheet yourself. See our AEO and GEO service or book a free audit.