Most GEO audits check your markup.This one asks the engines and counts the answers.
The two things called a GEO audit
Readiness checks
What most tools do- Schema markup completeness and validity
- Heading structure and answer-shaped content
- robots.txt access for AI crawlers
- Page speed, technical hygiene, internal links
- Output: a readiness score and a fix checklist
Visibility measurement
What this is- Presence rate — share of answers that name you at all
- Citation rate — share that cite your own domain as a source
- Share of voice — your mentions against every rival’s
- Average position — where in the answer you land
- The third-party pages the engines actually built the answer from
These are complements, not rivals. A readiness check is a hypothesis about why you might be invisible; a measurement tells you whether you are. Run the check by all means — but run it because a measurement said you had a problem, not instead of one.
Why readiness alone misleads
The uncomfortable finding in almost every audit we run is that the answer is not built from the brand's own site. It is built from a G2 category page, a Reddit thread, a comparison post on someone else's blog, and one vendor's documentation. You can hold perfect schema markup and lose every one of those answers, because the engine never needed your page.
That is why the most valuable section of the report is the list of source domains: the pages the answers were actually assembled from, ranked by how often each is cited across your prompt set. It converts “improve your content” into a list of specific URLs to go get onto, in order.
- A review site’s category page
- A community thread
- Someone else’s comparison post
- One vendor’s documentation
- Your site — only if the engine needed it
What one audit covers
~40 buyer-intent prompts, generated from your own site and then frozen. Four intent classes: category discovery, head-to-head comparison, problem-led search, and vendor-specific questions. Four engines. Three passes each. 480 sampled answers.
Three passes is not padding. A single answer from a language model is noisy — ask twice and you can get two different vendor lists. Sampling each prompt three times lets us report the spread alongside the number, so you know whether “41% presence” is a stable finding or a coin flip.
Freezing the prompt set is what makes a re-scan in 90 days a genuine before/after rather than two unrelated samples. It gets replayed verbatim. The full methodology is published — sample sizes, engines, dates, and the exact score formula.
Questions people ask before buying
- What is a GEO audit?
- GEO stands for generative engine optimization: getting named and cited inside AI-generated answers rather than ranking in a list of blue links. A GEO audit is supposed to tell you how you currently do on that. In practice the term covers two very different things — a check of whether your pages are technically readable by AI crawlers, and a measurement of what the engines actually say when someone asks about your category. Only the second one can tell you whether you have a problem.
- How is this different from a free GEO audit tool?
- Free tools generally analyse one site: its schema markup, headings, crawler access and content structure, and return a readiness score. That is useful and it is not visibility. A page can be perfectly structured and still never be mentioned, because the engine is citing a G2 category page and a Reddit thread instead. We measure the answers themselves — 40 buyer-intent prompts across four engines, three passes each, 480 sampled answers — and report how often you were named, how often your own domain was cited, and who was named instead.
- Why does the price sit between a free tool and an agency engagement?
- Because the work is automated. Agency GEO audits run from roughly €2,000 to €9,000 because a human reads the results and writes the deck. Ours runs the same measurement on a fixed pipeline and hands you the report, so the price is the compute plus a margin, not a consultant’s week. If you want a human to interpret it afterwards, that is a separate conversation and a separate price.
- Which engines do you measure?
- ChatGPT, Claude, Gemini and Perplexity, each through its own API with web search or grounding enabled. Engine-by-engine results are broken out, because disagreement between them is actionable: strong in Perplexity and absent from Gemini is a grounding problem, not a content problem.
- Do you measure the consumer ChatGPT interface?
- No, and we say so on every report. We measure through the providers’ APIs with grounding enabled. API answers are close to, but not byte-identical with, what a person sees in the consumer apps. Scraping those interfaces is brittle and against their terms; sampling the APIs is reproducible and auditable. We publish that tradeoff rather than hiding it.
- How long does it take?
- The free scan runs while you wait — a few minutes, because every answer is a live grounded engine call. The paid audit is promised within 24 hours and in practice completes in minutes — 480 grounded calls plus extraction, run through a queue.
Related
- AEO auditSame measurement, framed for answer engine optimization — and why a 48-point checklist is a hypothesis rather than a result.
- AI visibility auditWhat “visibility” means as a number: the four metrics, how they combine, and why a single score with no formula is unfalsifiable.
- The tool landscapePage checkers, answer-sampling platforms and agency engagements — what each category can and cannot tell you.
- MethodologyEngines, sampling, prompt construction, the score formula, and the limits we disclose rather than hide.
- The First-Cited AdvantageFor brand-new categories: why engines settle on one source per term, and the playbook for being it.
Find out what the engines actually say about you.
The free scan runs five prompts across all four engines and shows you the score, your closest competitors and the top cited sources. It takes a few minutes and it is enough to tell you whether the $349 audit is worth it.