The GEO Audit
A GEO audit measures whether AI search engines recommend you or your competitors. Here's what a real one covers, and how ours works.
Your buyer just asked ChatGPT who to hire. Were you in the answer?
Run the prompt yourself. Ask ChatGPT or Perplexity the question your customers ask before they buy. If a competitor comes back described like the obvious choice and you're nowhere, that conversation is repeating every day, at whatever scale your category gets searched. And you can't see any of it in Google Analytics.
What is a GEO audit?
A GEO audit (generative engine optimization audit) is a diagnostic that measures how often, and how accurately, AI search engines mention a brand when users ask category-relevant questions. It tests mention rate and citation rate across engines like ChatGPT, Perplexity, and Google AI Overviews, benchmarks share-of-voice against competitors, and identifies why AI answers cite some sites and ignore others. You'll also see it called an AI visibility audit, an LLM audit, or an AI search visibility audit. Same diagnostic, different label.
Where a traditional SEO audit measures rankings and links, a GEO audit measures whether AI answers recommend you. In 2026 that distinction decides deals, because you can hold your rankings and still lose the buyer who read an AI answer instead of a results page.
The GrindstoneSEO GEO Audit is a 30-day diagnostic across ChatGPT, Perplexity, Google AI Overviews and AI Mode, Gemini, Claude, and Copilot. It benchmarks your citation share against named competitors, finds the structural reasons AI answers skip you, and hands you a roadmap ordered by what moves the number fastest.
Your rankings held. Your AI visibility didn't.
The data on this is not subtle:
When an AI Overview shows up on a query, click-through to the organic results drops 58 to 61%. That's Seer Interactive's analysis of 25.1 million impressions across 42 clients over 15 months. A measured collapse, not a hot take.
Same study, other direction: brands cited inside the AI answer get 35% more organic clicks than brands that aren't. The answer box redistributes clicks toward whoever it cites.
AI Overviews now appear on roughly 15 to 25% of searches (Semrush's 10M+ keyword crawl and Conductor's 21.9M-search set land in the same band).
ChatGPT reports over 800 million weekly active users. That's a search engine with no rank tracker pointed at it.
Here's the part most people miss: AI citations are unstable. Passionfruit measured 68% of cited sources churning month to month across 11.2 million citations, and AirOps found only 30% of brands stay visible across back-to-back regenerations of the same answer. Bad news if you're winging it. Very good news if you're methodical, because the leaderboard resets constantly and most of your competitors aren't measuring at all.
I've been doing SEO since 2006. Every Google shift eventually had a playbook. This one does too, and it starts with knowing your actual baseline.
What the GEO audit measures: 12 checks across 3 layers
Measurement: what the engines say right now
Mention rate baseline.
How often each engine mentions you across a query set built from your real buyer questions, not keywords we made up.
Citation share-of-voice.
Your citation density against three named competitors on the same query set. This is the number your next quarter gets judged against.
Competitor citation forensics.
Which competitor wins which queries, and what they published or earned that you didn't.
Content and structure: what the engines can use
Entity representation.
How AI describes what you are, what you do, and who you serve. Getting described wrong is more common than being absent, and it poisons every answer you do appear in.
Intent alignment.
Pages that promise one thing and deliver another get skipped for citation because the engine can't tell what they're for.
Answer extractability.
Whether your key pages contain blocks an LLM can lift and cite without paraphrasing you into someone else's brand.
Query coverage.
Which branches of your buyers' question tree you cover and which you've left to competitors.
Dual-layer check.
Whether pages serve AI extraction and human conversion at the same time, or sacrifice one for the other.
Technical access: what the engines can reach
Crawler access.
robots.txt, CDN, and hosting rules that block GPTBot, ClaudeBot, PerplexityBot, or Google-Extended. Old wildcard rules kill AI visibility silently.
Schema and structure.
The markup layer that helps engines parse who you are and what each page answers.
Knowledge graph presence.
Wikidata, Wikipedia, LinkedIn, and the public entity layer LLMs lean on when they decide you're real.
Corroboration gaps.
The third-party surfaces citing your competitors and not you: review platforms, Reddit, YouTube, industry publications.
Every check produces a finding, every finding has a fix, and every fix is sequenced by impact. An inventory of problems is not a deliverable. An order of operations is.
Six engines, six different source graphs
Only about 12% of cited domains overlap between platforms (Superlines measured this across 15,000 queries). The pattern that wins ChatGPT does not automatically win Perplexity, and neither guarantees a Google AI Overview. Each engine pulls from a different source graph with different biases.
A single-engine scan is a sixth of an answer. Your buyers use all of them. The audit measures all of them, across seven answer surfaces (Google's AI Overviews and AI Mode are tested separately).
Free GEO audit tools vs a real diagnosis
You've seen the free GEO audit tools. We've run them on our own sites. They return a score, and the score is usually a prompt or two wrapped in a PDF. Fine as a curiosity. Useless as a diagnosis, because they can't tell you WHY you're invisible, what to fix first, or how you stack against the three competitors your sales team actually loses deals to.
Run the free scans. When you want the number explained and a plan attached to it, that's this.
GEO audit deliverables
Baseline visibility report.
Your mention rate and citation rate across all six engines, on a pinned query set you can re-test against forever. The pinned set matters: re-measuring with different queries is how agencies fake progress.
Competitor share-of-voice benchmark.
You against three named competitors, query by query. This is the page your leadership will screenshot.
12-check gap report.
Every finding scored by impact and effort.
90-day roadmap.
The fixes in the order they should happen, split into what your team can do, what your dev needs to do, and what has to be earned off-site.
The walkthrough call.
I present the findings, you ask questions, nobody reads a 40-page PDF cold.
Proof, not promises
We run this methodology on our own money sites first. House rule: nothing gets sold here that we haven't tested for science on our own domains. mclaughlincanyon.com sits #2 on its head term against AllTrails and federal-agency pages. aeneasvalley.net and fsbosandpoint.com sit #3 on theirs against national portals. All three verified on live SERPs, all three on near-zero link equity, because entity clarity and answer extractability do work that domain authority doesn't. The same methodology has run across client scans in finance, home services, real estate, and SaaS.
And yes, this site's own DR is low, and I've never built a link to raise it. Deliberately: I've kept this domain quiet since the Matt Cutts era, because staying off the radar was the play, and receipts matter more than badges. Judge the audit by the method and the sample report we'll show you on request.
Who the GEO audit is for (and who it isn't)
Good fit: businesses whose buyers research in ChatGPT, Perplexity, or Google AI before contacting anyone. Agencies who need a white-label GEO diagnostic for their clients. Marketing leads who need external benchmarking to defend a budget internally.
Bad fit: single-location local businesses (local SEO will return more per dollar), pre-launch startups with no entity footprint yet, and anyone shopping for a $99 scan. Those exist. Go run one, seriously. Come back when you want the why.
How the 30-day audit runs
Week 1: Discovery.
We map your category, build the pinned buyer-query set, and agree on the three competitors we benchmark against. You sign off on the query set before anything gets measured.
Weeks 2-3: Measurement and analysis.
All 12 checks run across all six engines. You get findings as they land, not at the end.
Week 4: Synthesis.
Gap report scored, roadmap sequenced, walkthrough call delivered.
After that, the roadmap is yours. Execute it in-house, hand it to your agency, or hire us to run it. Implementation is a separate engagement quoted after the audit, never bundled into it. Selling the prescription before the diagnosis is how this industry got its reputation.
GEO audit pricing
$3,500 flat.
Thirty days, six engines, three competitors, twelve checks, five deliverables. No tooling surcharges, no per-engine upsells, no "enterprise tier" that's the same work with a bigger invoice.
Book your audit kick-off →GEO audit FAQ
What is a GEO audit?
A GEO audit is a diagnostic that measures how often AI search engines like ChatGPT, Perplexity, and Google AI Overviews mention your brand when buyers ask category questions, and why they cite competitors instead. It covers mention rate, citation share-of-voice, entity representation, content extractability, and technical AI-crawler access.
How is a GEO audit different from an SEO audit?
An SEO audit measures rankings, links, and technical site health: the inputs to where you place on a results page. A GEO audit measures whether AI answers mention and recommend you. The disciplines overlap (Princeton's KDD research on generative engine optimization measured up to a 40% visibility lift from structural content changes), but the metrics, the source graphs, and the fixes are different. You need both.
Which AI engines does it cover?
Six engines across seven answer surfaces: ChatGPT (including ChatGPT Search), Perplexity, Microsoft Copilot, Claude, Gemini, and Google (AI Overviews and AI Mode tested separately). Cross-engine citation overlap is only about 12%, so single-engine scans miss most of the picture.
How long does a GEO audit take?
30 days from kick-off to walkthrough call. Week 1 is discovery, weeks 2-3 are measurement, week 4 is synthesis and delivery.
What does a GEO audit cost?
$3,500 flat, everything included. Implementation afterward is optional and quoted separately.
Can I do a GEO audit in-house?
You can measure one engine by hand today: ask it your buyer questions and log the answers. Doing it across six engines, on a repeatable pinned query set, with competitor benchmarking and a scored gap analysis is a tooling budget plus a senior practitioner-month. If you have both, go build it. Most teams have neither, which is why this is productized.
What happens after the audit?
You get the roadmap regardless of what you do next. If you want us to execute it, implementation is scoped as a separate engagement. If you'd rather run it in-house or through your existing agency, the roadmap is written to be handed off.
Related services
GEO Content Services: for executing the content side of your roadmap.
Link Building: for the off-site authority and corroboration gaps the audit surfaces.
AI Search Guides: the free playbook, including GEO vs SEO and the GEO checklist. Read how we think before you pay us to think for you.
Get the Newsletter
Tactical SEO, AI search testing, and the occasional rant. No fluff.