Published September 10, 2026 · 5 min read · By Adam Aksoy, Founder & Software Engineer
You measure GEO (Generative Engine Optimization) on four fronts: how often AI engines cite you for the questions your customers ask, how much traffic those engines send and how well it converts, how your site scores on a repeatable citability audit, and whether branded demand is rising. No single dashboard reports all four yet, so a working GEO program builds its own scorecard from those pieces and reviews it monthly. This guide is for the marketing lead who has to answer "is AI search doing anything for us?" with numbers rather than anecdotes.
Why can't I just watch rankings and organic traffic?
Because AI answers change what a ranking is worth. Pew Research Center tracked 900 U.S. adults across 68,879 Google searches in March 2025 and found that when an AI summary appeared, users clicked a traditional result on only 8% of visits, versus 15% when no summary appeared; 18% of all searches in the study produced one (Pew Research Center, 22 July 2025). Ahrefs reached a similar conclusion from the other direction: across 300,000 keywords, the presence of an AI Overview correlated with a 34.5% lower click-through rate for the top-ranking page (Ahrefs, 17 April 2025). A page can hold position one and still lose a third of its clicks. If rankings are your only metric, GEO progress is invisible and GEO losses look like a mystery.
Metric 1: AI citation share
AI citation share is the percentage of your target questions for which an AI engine names or links your site in its answer. It is the GEO equivalent of a rank tracker, and it is the leading indicator: citations move weeks before traffic does.
How to collect it: write down 20–40 questions your buyers actually ask (the ones your sales team hears, plus the "best X for Y" and "how much does X cost" queries). Ask each one in ChatGPT with search enabled, Perplexity, Google AI Mode or AI Overviews, and Gemini, on a fixed cadence — weekly is enough. Record three things per query and engine: were you cited (yes/no), which URL was cited, and which competitors were cited alongside you. Report it as a share: cited on 9 of 30 questions in Perplexity is a 30% citation share. Track the same question set every time; changing the questions resets the baseline.
What good looks like: on a new program the first target is simply moving from 0% to non-zero on the engines that already crawl you. From there, a healthy program grows citation share on the money questions (pricing, comparisons, "who does X in [city]") faster than on the generic ones, because those are the answers that end in a lead.
Metric 2: AI-referral traffic and its conversion rate
AI-referral traffic is the sessions that arrive from an AI engine's citation link, and the metric that matters is not the volume but the conversion rate. Semrush's July 2025 study found the average AI-search visitor is 4.4 times as valuable as the average organic-search visitor on conversion rate (Semrush, 21 July 2025), which matches what the Pew click data implies: the person who clicks through an AI answer has already read the summary and is arriving pre-qualified.
How to collect it in GA4: build a custom channel group called "AI search" with a rule on session source
matching chatgpt.com, chat.openai.com, perplexity.ai,
copilot.microsoft.com, gemini.google.com, and claude.ai, then compare
that channel's sessions, engagement rate, and key-event (conversion) rate against Organic Search. Two
cautions: Google AI Overviews and AI Mode clicks are reported inside Organic Search, not separately, so
this channel undercounts Google; and some engines strip the referrer, so treat the number as a floor. Even
as a floor, a channel that converts at several times the organic rate justifies the program on its own.
Metric 3: a scored, repeatable audit baseline
An audit score is the input metric: it measures whether your site is in a state an AI engine can crawl, understand, and quote. Citation share tells you the outcome; the audit tells you which lever to pull next. A useful audit scores the same items every time so the number is comparable month to month:
- Crawler access — GPTBot, ClaudeBot, PerplexityBot, and Google-Extended allowed in robots.txt and not blocked at the CDN.
- Entity signals — Organization and Person structured data with populated
sameAslinks, one canonical name and address everywhere. - Citability — the share of priority pages whose sections open with a one- or two-sentence quotable answer under a question-style heading.
- Sourcing — the share of statistical claims that are dated and deep-linked to a primary document, not a homepage.
- llms.txt — present, current, and pointing at the pages you most want cited (see what llms.txt is and how to write one).
Re-run the audit on the same schedule as the citation check. When the audit score rises and citation share follows a few weeks later, you have a causal story you can show a CFO.
Metric 4: branded demand
Branded demand is the volume of searches and direct visits for your name, and it captures the GEO value that never produces a click. Many AI answers name a company without linking it; the user then searches the brand or types the URL. Watch branded impressions and clicks in Google Search Console, Direct sessions in GA4, and, if you run them, brand-term impressions in Google Ads. A rising branded line alongside flat or falling generic organic clicks is the signature of AI answers doing the pre-selling for you.
What does a GEO scorecard look like at 30, 60, and 90 days?
| Checkpoint | What should have moved | What to fix if it has not |
|---|---|---|
| Day 30 | Audit score up; all four AI crawlers confirmed fetching in server logs | Crawler blocks, missing or broken structured data |
| Day 60 | First non-zero citation share on at least one engine; AI-search channel visible in GA4 | Priority pages still open with marketing copy instead of answers |
| Day 90 | Citation share growing on money questions; AI channel converting above organic; branded demand rising | Thin coverage — add answer-shaped pages for the uncited questions |
Which metric should I report to leadership?
Report citation share as the headline, AI-channel conversions as the business result, and the audit score as the explanation for both. Keep the raw question-by-engine log as the appendix; it is the evidence that turns "we are showing up in AI" into "we are cited on 14 of 30 buyer questions in Perplexity, up from 3." Progress should be measured against numbers, not asserted — the same standard we set out in our guide to what GEO is.
This scorecard is how NameVerse runs and reports every GEO engagement. If you want a baseline audit and a citation-share measurement for your own buyer questions, see our GEO services or get in touch.
Disclosure: the author is the founder of NameVerse, Inc., and this article discusses topics and services NameVerse offers commercially.
