GEO/AEOAI visibilityTool

GEO Audit: from technical expertise to market tool

Capiva has been optimizing for AI engines for over a year. That expertise became an open diagnostic tool: the GEO Audit. Any company can test whether AI cites their site — and know exactly what to fix.

Published on

77/100

Schema (JSON-LD)

97/100

Technical

89/100

Platforms

25/100

Citability
Next front

The opportunity

Optimizing for Google is SEO (Search Engine Optimization). Optimizing for ChatGPT, Gemini, Perplexity and Claude is GEO (Generative Engine Optimization). These are fundamentally different disciplines requiring different technical approaches. Most companies have invested years mastering the first while completely ignoring the second. According to BrightEdge research (2025), sites with FAQPage and Service schema combined receive 44% more citations from generative AI engines — yet fewer than 15% of enterprise sites have implemented the structured data that AI models need to cite them.

The pattern Capiva identified across multiple client engagements: sites with excellent content written for human readers that are functionally invisible to language models. The problem is never lack of content — it is lack of structure. Content without schema markup, without fact-dense passages in the 134 to 167 word optimal extraction range, and without the statistical density that AI models use to assess source reliability simply does not get cited regardless of its quality.

Missing schema markup means AI cannot categorize the page content. Absent fact-dense passages mean no extractable text blocks in the optimal range. Client-side rendering delivers empty HTML to AI crawlers. These are technical problems that no human visitor notices — but they determine whether AI engines cite a site or ignore it entirely.

What we built

The GEO Audit evaluates any website across 6 AI visibility dimensions: Schema Markup scoring JSON-LD implementation, AI Citability measuring passage quality and extraction readiness, Brand Authority assessing mention frequency across AI platforms, Content E-E-A-T evaluating experience, expertise, authoritativeness and trustworthiness signals, Technical Foundations checking rendering, speed and crawlability, and Platform Readiness measuring presence across ChatGPT, Gemini, Perplexity and Claude.

The methodology is backed by peer-reviewed research from BrightEdge, Frase.io, Cloudflare and Stackmatix published in 2025. The tool operates with 5 parallel AI agents that complete a full diagnostic in approximately 60 seconds. Development included 30 unit tests validating scoring consistency and over 7 sites audited during the calibration phase. The scoring algorithm weights each dimension based on its demonstrated impact on AI citation frequency.

The strategic decision was to productize as an open diagnostic tool rather than a closed consulting service. Anyone can access the GEO Audit, enter a URL and receive a complete diagnostic with overall score, identified issues organized by severity and a prioritized remediation roadmap. This approach serves as a technical demonstration of Capiva's AI visibility expertise before any sales conversation occurs.

Lead generation follows naturally from the diagnostic experience: a company that discovers their site scores an F grade across AI visibility dimensions wants to understand how to reach B+ or higher. The tool demonstrates concrete technical expertise rather than making abstract capability claims, converting diagnostic users into consulting clients through demonstrated value.

Results and evolution

The capiva.tech website serves as the tool's continuous test case and living proof of the methodology. After implementing the GEO Audit's own recommendations: Schema Markup increased to 77 out of 100, Technical Foundations reached 97 out of 100, and Platform Readiness scored 89 out of 100. These improvements were achieved through structured data implementation, server-side rendering optimization and fact-dense content restructuring.

Citability at 25 out of 100 and Content E-E-A-T are the active optimization fronts. Every improvement on Capiva's own site feeds directly into the recommendation roadmap for other sites — the consultancy operates as its own most demanding test case, ensuring recommendations are validated in production before being applied to client sites.

The GEO Audit is live at capiva.tech/diagnostico. Any URL can be audited without registration for the basic score. The premium report includes detailed remediation steps, competitor benchmarking and AI citation probability estimates across ChatGPT, Gemini, Perplexity and Google AI Search.

Factual summary

Capiva developed the GEO Audit, a visibility diagnostic tool for generative AI engines such as ChatGPT, Gemini, Perplexity and Claude. The tool evaluates sites across 6 dimensions: Schema Markup, AI Citability, Brand Authority, Content E-E-A-T, Technical Foundations and Platform Readiness. The methodology is backed by research from BrightEdge, Frase.io, Cloudflare and Stackmatix published in 2025. According to BrightEdge research, sites with FAQPage and Service schema combined receive 44% more citations from generative AI. The tool operates with 5 parallel agents, 30 unit tests validating scoring consistency, and generates a complete diagnostic in approximately 60 seconds. The GEO Audit is open: any URL can be audited without registration for the basic score. Capiva uses the tool on its own site as a continuous test case, with Schema at 77 out of 100 and Technical at 97 out of 100. The tool is available at capiva.tech/diagnostico.

What this means for you

If an AI company needs to optimize for AI, imagine those who don't work with it every day. The question isn't whether your site is visible to AI. It's when your competitors will find out before you do.

Next step

Test your site now

Enter the URL and get the full diagnostic in 60 seconds.

Test my site

Or reach out directly:

WhatsApp

A GEO audit checks whether answer engines cite your brand when they answer a question in your market. It has two layers. The structural layer measures whether your page is extractable: self-contained passages, question-and-answer blocks, explicit data. The outcome layer asks the engine the question and counts how many times your name appears in the answer. Only the second is a result. The first is a proxy.

Why is scoring the page not enough?

Ranking and being cited are independent. You can verify this on your own domain. The test costs cents.

We ran the measurement on capiva.tech. One of our pages ranks first on Google for the definitional query of its topic. Six questions across three engines, eighteen observations. Number of times the brand was named in the answer: zero. One engine did retrieve the page and answered without saying whose text it was. The other two retrieved nothing of ours.

Being a source is not being cited. The passage it took was a clean definition. And anonymous.

This is not a ranking problem. The ranking is at the top.

Proxy or outcome: what is the practical difference?

Structural layer (proxy)Outcome layer
What it measureswhether the page is extractablewhether the engine names you
How it gets itparses your HTMLasks the engine and reads the answer
Costzero, runs in secondscents per question, per engine
Answersis my content ready?do I show up?
Typical errorhigh score and zero citationsnone, it is the result

Both are useful. The structural one is diagnostics. You work on it daily. The outcome one is the scoreboard, and a scoreboard does not need daily reading: an engine's answer does not change overnight. Weekly measurement already captures the variation and avoids paying seven times for the same answer.

What if the audit reports zero citations?

Then you have an honest number. Better than a score of 80 that means nothing.

Zero citations with good rankings points at entity authority. Not at writing quality. Rewriting the paragraph does not move that number. What moves it is your brand existing as a resolvable entity, and the citable statement carrying your name inside it, without turning into an ad.

Frequently asked questions about GEO audits

What is a GEO audit?

A GEO audit is the check of whether answer engines cite a brand when answering questions in a market. The audit combines two measurements: the structural one, which analyses whether a page is extractable by a model, and the outcome one, which sends the question to the engine and counts brand mentions.

What is the difference between a GEO audit and an SEO audit?

An SEO audit measures whether a page ranks in a list of links. A GEO audit measures whether a brand is named inside a generated answer. These are independent outcomes: a page can hold the first position in traditional search and be mentioned in no generative answer at all.

How much does a GEO audit cost?

The cost of a GEO audit splits into two layers. The structural layer costs close to nothing, because it only parses the page HTML. The outcome layer costs per question and per engine, in the range of cents per query. The total depends on how many questions are monitored and how often.

How often should a GEO audit be repeated?

The ideal cadence for a GEO audit is daily on the structural layer and weekly on the outcome layer. The structural layer carries no per-run cost. An engine answer does not change from one day to the next, so daily outcome measurement pays several times for the same information.

Which engines should a GEO audit query?

A GEO audit should query at minimum the three engines with the largest conversational search usage: ChatGPT, Gemini and Perplexity. The three behave differently. Some retrieve pages in real time before answering, while others answer from model training without consulting the web for that particular question.

Does ranking first on Google guarantee an AI citation?

Ranking first on Google does not guarantee an AI citation. Traditional search and generative engines are independent selection systems. A page can hold first position and appear in no generative answer, because the engine may answer from model knowledge without retrieving any page for that query.

What is citability in a GEO audit?

Citability is how far a passage of a page can be extracted and reused by an engine without depending on surrounding context. Self-contained passages, direct answers placed right below a question, and explicit data raise citability. Paragraphs that depend on the previous paragraph lower citability.

Is being retrieved the same as being cited?

Being retrieved is not the same as being cited. Retrieval means the engine fetched and read the page before answering. Citation means the engine named the brand in the answer delivered to the user. A site can be retrieved and not cited, when the engine uses information without attributing it.

What is AEO and how does it relate to GEO?

AEO stands for Answer Engine Optimization and deals with appearing in direct answers. GEO stands for Generative Engine Optimization and deals with appearing in answers generated by language models. In practice both terms describe the same goal: being the named source of an answer.

Does an FAQ block help in a GEO audit?

An FAQ block helps and ranks among the highest-effect levers in a GEO audit. Visible questions and answers create short, self-contained passages already shaped the way an engine needs to answer. Each question and answer pair becomes an independent candidate for citation.

What is the ideal answer length to be cited?

The ideal length for a citable answer sits between forty and sixty words. Below that band there is not enough context for the passage to stand alone. Above it the engine has to cut, and the cut usually breaks the very sentence that carried the source attribution.

Does promotional language hurt citation?

Promotional language hurts citation. A passage that presents itself as advertising is a passage the engine avoids reproducing as fact. Neutral, verifiable definitions are reused far more often than sentences praising a product or promising a result to the reader.

Is structured data enough to be cited?

Structured data alone is not enough to guarantee citation. Markup helps a machine understand the content type of a page, but citation depends on the visible text being extractable and the brand being recognisable as an entity. Markup without citable text produces no mention at all.

What should you do when an audit reports zero citations?

Zero citations should be treated as a diagnosis of entity authority, not of writing quality. It is worth checking whether the brand exists as a recognisable entity in public databases, whether the citable statement carries the name inside it, and whether the measured questions are the market questions.

How do you know a GEO audit is trustworthy?

A trustworthy GEO audit delivers the raw engine answer alongside the report. An audit that reports a score without showing the answer that produced that score is not auditable. It is worth asking for the list of questions used as well, because question choice determines the final result.

Does a GEO audit replace an SEO audit?

A GEO audit does not replace an SEO audit. The two measure different stages of the same journey. Without indexing a page is not retrievable by any engine, so the technical search audit remains a prerequisite for the GEO audit rather than an alternative to it.

← Back to Cases