How Do You Measure GEO and AEO?

1. AI share of voice (citation share)

This is the headline GEO metric. AI share of voice is the percentage of AI answers, for a defined set of category questions, that mention or cite your brand — measured against your competitors. If you run 50 buyer questions through ChatGPT and your brand appears in 12 answers, your citation share for that engine is 24%.

Track it per engine — ChatGPT, Perplexity, Gemini, Google AI Overviews, Copilot — because, as the Profound data shows, they cite different corners of the web. A single blended number hides the engine where you are invisible. This is the closest thing GEO has to a "ranking," and it is the number a serious generative engine optimization program moves month over month.

2. Prompt-set monitoring

You cannot measure share of voice without a stable prompt set — a fixed list of the real questions your buyers ask, run on a schedule so the results are comparable over time. A good prompt set covers three intents: category questions ("best AI SEO agency for SaaS"), comparison questions ("Agency A vs Agency B"), and branded questions ("is [your brand] any good"). For each run you log whether you were mentioned, whether you were cited with a link, your position in the answer, and which competitors appeared. Re-run it monthly. The change in that table is your progress.

3. AI-referral traffic in your analytics

When someone does click through from an AI answer, that visit lands in your analytics — but GA4 files it under generic "Referral" or, worse, "Direct," so it stays invisible by default. The fix is a custom channel group with a regex on the traffic source matching chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com, and similar hosts, placed above the built-in Referral rule so GA4 checks it first. Once it is live, AI assistants show up as their own traffic row you can trend and attribute to conversions.

This metric matters disproportionately because AI traffic is high-intent: Semrush reports the average AI search visitor is worth 4.4x a traditional organic visitor, and AI search traffic grew 527% year over year (Semrush). A small AI referral number today can carry outsized revenue.

4. Answer-box and AI Overview ownership

This is the primary answer engine optimization metric. It tracks how often you own the direct answer — the featured snippet, the AI Overview citation, the "people also ask" expansion, the voice result — for your target questions. Practically, you monitor: does an AI Overview appear for this query at all, and if so, are you one of the cited sources? Because AI Overview citations now diverge sharply from top-10 rankings (Ahrefs), you have to check the box itself, not infer it from your position. Track presence, citation, and the exact passage the engine lifted.

5. Brand-mention volume and sentiment

Not every AI reference is a clickable citation. Engines increasingly name brands in prose — "firms like X and Y" — without a link. Brand-mention volume counts how often you are named across AI answers, cited or not, and sentiment captures whether the mention is favorable, neutral, or a warning. This metric is the leading indicator: mentions usually rise before citations and referral traffic do, because being talked about is what teaches an engine you are a real entity worth quoting. Watching mentions climb is often the first proof your work to get cited by AI is landing.

You have three layers, and a real program uses all three.

1. Manual spot-checks (free, start here). Open ChatGPT, Perplexity, Gemini, and a Google AI Overview query in a fresh/incognito session, run your top 10 buyer questions, and log the answers in a spreadsheet: mentioned? cited? competitors named? This is tedious but honest, and it is exactly how a free AI visibility audit baselines where you stand today. Do it once and you will know more than most of your competitors.

2. Analytics (free, already yours). Set up the GA4 custom channel group above so AI-referral traffic and its conversions are visible. This is the one metric you own end to end, with real revenue attached — do not skip it.

3. Dedicated AI-visibility platforms (paid, for scale). Tools built for this — the category includes Profound, Semrush's AI toolkit, and several newer trackers — run large prompt sets on a schedule across every engine and chart your share of voice for you. They turn a monthly afternoon of manual checks into a live dashboard. Our own AI visibility checker is the fast, no-login version of that first look.

The sequencing that works: baseline manually, wire up GA4 immediately, then automate the prompt set once you know which questions matter. Measurement is the whole point of AI search optimization — you cannot improve a citation rate you never recorded.

Be skeptical of universal benchmarks; a "good" AI share of voice depends entirely on your category and competition. Judge yourself against your own last month and your named competitors, not an industry average someone invented. Movement is the metric: is your citation share rising, are you appearing in engines that ignored you before, is AI-referral traffic growing?

And ignore vanity signals. The clearest cautionary tale is llms.txt: Ahrefs studied 137,210 domains and found 97% of llms.txt files received zero requests in a single month (Ahrefs, June 2026). A file you installed is not a metric; a citation an engine actually made is. Measure outcomes — mentions, citations, share of voice, referral revenue — never checkboxes.

"Rankings tell you where you sit in a list nobody reads anymore. The metrics for GEO and AEO tell you whether the machine writing the answer knows your name. Baseline it, do the work, and show the movement every month — that is the whole discipline." — Thomas, Founder of AISEO USA, 16 years in digital marketing