GEO Measurement
GEO can be measured.
Here's exactly how we do it.
Updated
Most teams can't say whether they're winning or losing in AI search, because they're measuring the wrong things. We measure four, every 30 days, across every engine your buyers use.
Methodology · Last updated
Free 30-minute call. A working conversation, not a pitch.
GEO measurement answers one question: when your buyers ask AI which vendor to choose, do you show up, and are you the answer? Resonate Labs scores that across a fixed set of buyer-intent queries, run monthly, on every engine that matters.
A fixed query set
150 buyer-intent queries built from your category, personas, and competitive landscape. The same set runs every cycle, so movement is comparable month over month.
Four metrics, four engines
Visibility, citation, share of voice, and win rate, scored separately on Gemini, Claude, ChatGPT, and Perplexity, because each engine cites differently.
Leading and lagging signals
Crawler traffic moves first and citation visibility follows. We track both, so the program shows results before the headline number catches up.
What we measure
Four metrics, each answering a different question. We score every one separately, on every engine.
Visibility
Are you in the answer at all?
The share of tracked queries where an engine mentions you. Being mentioned is the floor; it doesn't mean you were the recommendation.
Citation
Is your page the source?
The share of responses where an engine references one of your pages as its source. This is what on-page content work moves most directly.
Share of voice
How much of the conversation is yours?
Your portion of all vendor mentions across the full query set, measured against the competitors buyers are comparing you to.
Win rate
When you appear, do you win?
Of the queries where you're visible, the share where you're the primary recommendation: the vendor the engine surfaces first.
How the measurement runs
The same sequence every cycle, from the foundation that builds the query set to the monthly score across four engines.
-
01
Foundation Review · once
Build the model the engines see
A knowledge graph of your category, ideal customer profiles, positioning, and competitors. Every query in the audit comes from here.
-
02
Query set · generated
Turn the graph into buyer questions
150 buyer-intent queries spanning the buying stages, from problem identification through vendor shortlisting.
-
03
Monthly run · every 30 days
Score across four engines
The full set runs against Gemini, Claude, ChatGPT, and Perplexity. We score visibility, citation, share of voice, and win rate on each.
-
04
Per-platform read · every cycle
See where you're strong and where you're absent
Results break down by engine. A strong score on one and a blank on another is a content gap you can target.
-
05
Refresh · quarterly
Keep the set honest
The Foundation Review refreshes every quarter, so the queries track how buyers search this month, not last quarter.
What we test on ChatGPT, Claude, Gemini, and Perplexity
Resonate Labs runs the same test on all four engines, so a score on one can be compared with a score on another. Each engine gets the full set of 150 buyer-intent queries every 30 days, and each is scored on its own.
| What's tested | ChatGPT | Claude | Gemini | Perplexity |
|---|---|---|---|---|
| Query volume | ChatGPT answers all 150 buyer-intent queries in the set, from problem identification through vendor shortlisting. | Claude answers all 150 buyer-intent queries in the set, from problem identification through vendor shortlisting. | Gemini answers all 150 buyer-intent queries in the set, from problem identification through vendor shortlisting. | Perplexity answers all 150 buyer-intent queries in the set, from problem identification through vendor shortlisting. |
| Cadence | ChatGPT is re-run every 30 days on the same query set. | Claude is re-run every 30 days on the same query set. | Gemini is re-run every 30 days on the same query set. | Perplexity is re-run every 30 days on the same query set. |
| Scoring method | Visibility, citation, share of voice, and win rate, scored for ChatGPT alone. | Visibility, citation, share of voice, and win rate, scored for Claude alone. | Visibility, citation, share of voice, and win rate, scored for Gemini alone. | Visibility, citation, share of voice, and win rate, scored for Perplexity alone. |
| Signals captured | Whether ChatGPT names you, which of your pages it cites, which competitors it names instead, and whether you're its first recommendation. | Whether Claude names you, which of your pages it cites, which competitors it names instead, and whether you're its first recommendation. | Whether Gemini names you, which of your pages it cites, which competitors it names instead, and whether you're its first recommendation. | Whether Perplexity names you, which of your pages it cites, which competitors it names instead, and whether you're its first recommendation. |
Why testing one AI engine misses most of the answer
A score from one engine tells you very little about the other three, because the engines don't cite the same sources. The overlap in what ChatGPT and Perplexity cite is only about 11%, so roughly 89% of citation opportunities are specific to a single engine (Growth Unhinged). A team that tests only ChatGPT can report a strong number while it's absent from Gemini or Perplexity on the same buyer question. That's why Resonate Labs runs the same 150 queries on ChatGPT, Claude, Gemini, and Perplexity and scores each one separately.
Most single-surface setups cover part of the picture. Google Search Console's generative-AI reports show impressions inside Google's AI features, but only for Google. The AI module in the Semrush toolkit many B2B teams already pay for tracks ChatGPT and Google AI Overviews. Both are useful starts. Neither tells you whether Claude or Perplexity names you when your buyers ask them which vendor to choose.
The per-engine gaps are also where the work comes from. A strong score on one engine next to a blank on another points at a specific content gap you can target next month. Platform divergence covers how each engine's sources differ.
Multi-engine GEO measurement: Resonate Labs, Profound, and Omniscient Digital
Buyers shortlisting a GEO agency or AI visibility tool usually end up comparing three models: a tool that tracks, an agency that writes, and a service that measures and hands back the fixes. Resonate Labs is the third: a published, fixed test across ChatGPT, Claude, Gemini, and Perplexity, with a prioritized action plan. Profound and Omniscient Digital are the better fit for some teams, and the last column says when.
| Provider | What it is | How it measures AI visibility | What you get back | Choose it when |
|---|---|---|---|---|
| Resonate Labs | A GEO (generative engine optimization) service for B2B brands. | The same 150 buyer-intent queries on ChatGPT, Claude, Gemini, and Perplexity every 30 days, four metrics scored per engine, with the methodology published on this page. | Per-engine scores, the competitors named instead of you, and a prioritized action plan with the exact fixes, including crawler and rendering blockers. | You want to know where you stand on all four engines and which pages to publish next. |
| Profound | An AI visibility measurement platform (software). | Continuous multi-engine tracking, deeper and more frequent than a monthly run. | Dashboards and content briefs. It does not diagnose technical blockers such as AI crawlers getting an empty page. | You need always-on monitoring with alerts, and your team has capacity to act on what it shows. |
| Omniscient Digital | A full-service B2B SaaS content and SEO agency that sells GEO inside a content program. | GEO is measured as part of a revenue-framed content retainer. | Content written and shipped for you. | You want one partner to write and ship everything, and you have budget for a monthly retainer. |
For the wider field, including Peec AI and other trackers, see GEO Agencies & Tools, Compared.
Two signals, two speeds.
Not every metric moves at the same pace, and confusing the two is how a working program gets misread as a failing one. AI crawler traffic is the leading indicator. When new content ships, the engines pull it within days, and you see it in your logs before anything changes in the answers.
Citation visibility is the lagging indicator. It moves once the engines have indexed, weighed, and started citing the new content, which takes a cycle. We report both, so the early signal stays visible while the headline number catches up.
What the method surfaced on a live engagement.
On one active B2B SaaS engagement, the method tracked the two signals moving in sequence. Crawler traffic lifted first as new pages were pulled, then citation visibility followed as the engines indexed them. The full breakdown of pages shipped, crawler lift, persona visibility, and citation visibility is in the Insynctive case study.
Reading movement, not a moment
A single measurement is a data point, not a position. Run the same query against the same engine twice in a week and the wording, and sometimes the vendors named, will shift, because these are probabilistic systems. The size of that swing is documented: asked the same question 100 times, ChatGPT or Google's AI return the same list of brands less than 1% of the time (SparkToro and Gumshoe, 2,961 runs), and roughly 40 to 60% of the domains an engine cites turn over from one month to the next (Profound). That is exactly why the same 150-query set runs every 30 days against the same four engines, scored the same way: holding the ruler fixed turns single readings into a trend line, and the trend line is what tells you whether the program is working.
So the number that matters is not any one cycle's score; it is the direction and the slope across cycles. A real gain shows up as sustained movement over multiple cycles and across more than one engine, not a one-month spike that a rerun would erase. When the only thing changing is your content and the web around it, month-over-month movement is signal rather than variance. The GEO White Paper covers why a single snapshot is variance, not position, and how the fixed baseline separates the two.
Connecting visibility to pipeline
Here we are deliberately honest, because the category is full of vendors who aren't. The chain from AI visibility to closed revenue is real, but it is not yet deterministic, and we say so plainly rather than sell a number we can't defend. No independent study has traced the full chain from GEO investment to visibility to pipeline to revenue, and AI-driven clicks routinely land in analytics as direct or organic with no referrer, because Google gives no native way to isolate a click from an AI Overview. The limits of GEO is our full accounting of what's still unsolved, and clean revenue attribution is on that list.
What we can measure directly are the leading indicators. AI crawlers hit your logs within days of new content shipping. On the Google side, Search Console's generative-AI performance reports now surface the impressions your pages earn inside AI features, alongside your organic impressions and clicks, so the visibility that used to be invisible is becoming a metric you can watch for free. And AI referral traffic, the sessions arriving from ChatGPT, Perplexity, and the rest, shows up in analytics and tends to convert well, because the visitor arrived pre-qualified by the answer that sent them: in one large ecommerce dataset, ChatGPT referrals converted about 31% better than non-branded organic search (Visibility Labs, via Search Engine Land), a directional read for B2B rather than a measured one.
What stays directional is the last hop to a booked deal. AI-influenced pipeline rarely carries a clean source tag, because the buyer built their shortlist inside a chatbot weeks before they ever filled out a form. So we triangulate instead of fabricate: the leading indicators above, the correlation between visibility gains and pipeline movement over the same window, and self-reported "how did you first hear about us" on inbound. Honest triangulation that holds up in a board meeting beats a precise attributed figure that falls apart the moment someone asks how it was calculated.
The KPIs that connect visibility to pipeline
Resonate Labs reads seven KPIs in sequence, from the first sign an engine has seen your content to the buyer telling you where they heard of you. Visibility and share of voice sit in the middle of that chain. Read together over the same window as your pipeline, they show whether the program is moving buyers, without claiming a dollar figure nobody can defend.
| KPI | What it shows | Where it's measured | When it moves |
|---|---|---|---|
| AI crawler hits | The engines are pulling your new pages. | Your server logs. | Within days of new content shipping. |
| AI-feature impressions | Google is showing your pages inside its AI features on Search. | Google Search Console's generative-AI performance reports. | Once Google has indexed the new pages. |
| Visibility rate, per engine | The share of the 150 buyer-intent queries where each engine names you. | The monthly Resonate Labs run on ChatGPT, Claude, Gemini, and Perplexity. | Typically within a measurement cycle. |
| Share of voice, per engine | Your portion of vendor mentions against the named competitors buyers compare you to. | The monthly Resonate Labs run on ChatGPT, Claude, Gemini, and Perplexity. | As a trend across multiple 30-day cycles. |
| Win rate on high-intent queries | How often you're the first recommendation when the buyer is closest to choosing. | The shortlisting queries in the monthly run. | As a trend across multiple 30-day cycles. |
| AI referral sessions and conversion | Visitors arriving from ChatGPT, Perplexity, and other assistants, and how they convert. | Your web analytics. | As visibility and citation rise. |
| Self-reported source | Buyers naming an AI assistant as where they first heard of you. | A "how did you first hear about us" field on inbound. | With each new inbound lead. |
Putting it in front of a CFO
The metrics that travel up to a CFO or board are not the operational ones. Four things carry: share of voice against the named competitors, because leadership understands winning or losing the category conversation; the trend line, because direction over cycles is the proof the program works; win rate on the highest-intent queries, because showing up when the buyer is closest to choosing is what ties to revenue; and the leading indicators tied to traffic, because they move first and show early progress. The per-query, per-engine detail stays at the working level, where it drives the next month's content.
Two traps sink these reports. The first is collapsing everything into a single "AI visibility score." It hides the per-engine reality that a strong ChatGPT number can sit right next to a blank on Google, and it is precisely the per-engine gaps that tell you what to fix. Platform divergence is why the score has to stay split. The second is reporting a raw number with no baseline, which any CFO reads as noise. Report movement against the fixed baseline, per engine, with the competitive frame, and you have a dashboard that survives a budget review instead of raising more questions than it answers.
Each metric, precisely defined
Each metric has a precise definition, and a few are easy to conflate. Visibility isn't citation, and neither is share of voice. We keep the definitions, with examples, in one place: the GEO Metrics Glossary.
Frequently asked questions
How often does Resonate Labs measure GEO results?
Every 30 days. The full query set runs monthly against all four engines, so month-over-month movement is comparable. The Foundation Review that generates the query set refreshes quarterly.
What's the difference between visibility and citation?
Visibility means an engine mentioned you in its answer. Citation means the engine used one of your pages as its source. You can be visible without being cited, and citation is what on-page content work moves most directly.
Which AI engines does Resonate Labs measure?
Gemini, Claude, ChatGPT, and Perplexity, scored separately, because each engine has its own source diet and cites differently.
How long before GEO results show up?
Crawler traffic moves first, often within days of shipping new content. Citation visibility is the lagging indicator and typically moves within a measurement cycle, once the engines have indexed and started citing the new pages.
Can GEO results actually be measured, or is it guesswork?
They can be measured. Resonate Labs scores visibility, citation, share of voice, and win rate on a fixed set of 150 buyer-intent queries, so change is tracked against a stable baseline rather than estimated.
How do you track AI visibility over time?
By holding the ruler fixed. The same 150-query set runs every 30 days against the same four engines, scored the same way, so single readings become a trend line. Because these are probabilistic systems, any one cycle carries variance, so what matters is the direction and slope across multiple cycles and more than one engine, not a single month's number.
How do you connect AI visibility to pipeline or revenue?
Honestly, and without overclaiming. We measure the leading indicators directly: AI crawler hits in your logs, the impressions Google Search Console now reports for AI features, and AI referral traffic that tends to convert well. The last hop to a booked deal stays directional, because buyers build their shortlist inside a chatbot weeks before they fill out a form, so AI-influenced pipeline rarely carries a clean source tag. We triangulate with those indicators, the correlation between visibility and pipeline over the same window, and self-reported sourcing, rather than claim an attributed dollar figure we can't defend. Deterministic revenue attribution is still unsolved, and we say so.
How do I report AI visibility to my CFO or board?
Lead with the four metrics that travel up: share of voice against named competitors, the trend line over cycles, win rate on the highest-intent queries, and the leading indicators tied to traffic. Keep per-query, per-engine detail at the working level. Avoid two traps: collapsing everything into a single "AI visibility score," which hides the per-engine reality, and reporting a raw number with no baseline, which reads as noise. Report movement against the fixed baseline, per engine, with the competitive frame.
Can I measure AI visibility in Google Search Console?
Partly, and it's a useful free start. Google Search Console's generative-AI performance reports now surface the impressions your pages earn inside AI features on Search, alongside your organic impressions and clicks. It's a real leading indicator for the Google surface, though it doesn't yet include click data for AI features and it only covers Google, not ChatGPT, Claude, or Perplexity. We track it alongside the per-engine citation scoring that covers the engines Search Console can't see.
Do we need to measure all four engines, or is one enough?
All four, because they don't agree. The overlap in what ChatGPT and Perplexity cite is only about 11%, so roughly 89% of citation opportunities are specific to a single engine (Growth Unhinged). A strong score on ChatGPT tells you almost nothing about whether you're visible on Gemini or Perplexity, where your buyers may be asking the same question. We score each engine separately for that reason, and the per-engine gaps are where the next month's content work comes from.
Does showing up in AI answers actually drive pipeline?
It contributes, and it can be tracked, but not yet attributed to the dollar. AI referral traffic tends to convert well because the answer pre-qualifies the visitor: in one large ecommerce dataset, ChatGPT referrals converted about 31% better than non-branded organic search (Visibility Labs, via Search Engine Land). Resonate Labs tracks the KPIs in between (crawler hits, AI-feature impressions, per-engine visibility and share of voice, win rate, AI referrals, and self-reported source) and reads them against pipeline over the same window rather than claim a causal revenue figure.
No card required
See where you stand on all four.
Start with a free Foundation, the map of your buyers, your competitors, and what AI can read on your site. The full AI Visibility Crawl is where the real measurement happens, scored across every engine and every metric:
- Which queries your buyers actually ask AI
- Where you're visible, cited, or absent today
- What the first 30 days would move