Diagnosis & AI Visibility — Measuring the Status Quo

Lesson 2.1: What an AI Visibility Score Measures

An AI Visibility Score is a composite KPI metric (typically 0–100) that combines three dimensions into a single metric:

1. Technical Machine Readability: Can the AI find and correctly read the brand's facts at all? (robots.txt, JSON-LD, llms.txt)
2. Content Accuracy (Brand Accuracy): Do the answers that AI models generate about the brand match reality?
3. Recommendation Rate (Citation Rate): Is the brand actually recommended and cited as a source by AI models for relevant industry questions?

A common mistake is measuring only one of these dimensions. A company can be technically perfect (Dimension 1: 95/100), but the AI still recommends the competitor because their content is more current or better structured (Dimension 3: 20/100).

Score Interpretation

Score	Grade	Meaning
90–100	A+	Excellent — Brand is AI-ready, actively recommended
80–89	A	Very good — minimal gaps, quick to close
70–79	B+	Good — basic structure exists, but context signals missing
60–69	B	Solid basis — significant gaps in multiple categories
40–59	C	Below average — AIs often recommend the competitor
20–39	D	Poor — largely invisible to AI
0–19	F	Critical — no semantic infrastructure

Important for client communication: The score is not a pure "code score." A score of 55 doesn't mean "the website has a few technical issues" — it means the brand is practically invisible for AI-driven purchase decisions.

Lesson 2.2: The 6 Evaluation Categories

A professional GEO audit is structured into 6 categories. Each illuminates a different aspect of AI visibility.

Category 1: Schema & Structured Data (Weight: High)

JSON-LD Schema is the "language" in which AI models understand a brand most reliably. Without structured data, the AI has to guess and frequently invents facts (hallucination).

Finding	Impact	Frequency
❌ No Organization schema	AI doesn't know the brand as an entity	~40% SMEs
❌ FAQ schema missing	Massive citation potential wasted	~65%
❌ Person schema missing	No E-E-A-T signals for experts	~80%
⚠️ Schema errors	AI adopts errors 1:1	~30%
✅ Complete schema set	Excellent basis for AI citations	~5%!

Analogy: Imagine applying for a job, but your resume contains neither your name nor your work experience — just a vague "I'm dynamic and a team player." That's exactly how a website without structured data looks to an AI.

Category 2: Discovery & Crawlability (Weight: High)

Before an AI can cite a brand, it must be able to find and read the content. Checkpoints: robots.txt, llms.txt, sitemap.xml, bot blocking at the network level.

The Anti-Bot Paradox: Companies block all bots for IT security reasons and thereby unintentionally lock out the AI agents that are becoming the primary information source. A robots.txt that blocks GPTBot is the digital equivalent of denying journalists access to your own press conference.

Category 3: Brand Accuracy (Weight: Medium)

The most impactful category for management presentations: "What does the AI say RIGHT NOW about our brand?" When ChatGPT quotes wrong prices or recommends the main competitor — immediate urgency to act arises.

Category 4: E-E-A-T Signals (Weight: Medium)

Experience, Expertise, Authoritativeness, Trustworthiness — the universal trust signal for all AI models. Checkpoints: Person schema, sameAs links, Wikipedia/Wikidata entries, awards and certifications.

Category 5: Technical Foundation (Weight: Medium)

Overlaps with traditional Technical SEO: HTTPS, Core Web Vitals, mobile optimization, heading hierarchy, semantic HTML structure.

Category 6: Competitive Positioning (Weight: Low)

The relative position to the competition determines whether an AI recommends your brand or another. Benchmarking against 1-5 competitors.

Lesson 2.3: The 16+ Individual Checks

Within the 6 categories, 16+ reproducible checks are performed — your audit checklist for every GEO project:

#	Check	Category
1	robots.txt AI crawler access	Discovery
2	llms.txt existence & quality	Discovery
3	sitemap.xml reachability	Discovery
4	JSON-LD Organization Schema	Schema
5	JSON-LD FAQPage Schema	Schema
6	JSON-LD Product/Service Schema	Schema
7	JSON-LD Person Schema	E-E-A-T
8	JSON-LD BreadcrumbList	Schema
9	Semantic context markers	Schema
10	Meta tag quality	Technical
11	Heading hierarchy	Technical
12	Alt-text quality (Visual GEO)	Technical
13	E-E-A-T signal detection	E-E-A-T
14	Wikipedia/Wikidata presence	E-E-A-T
15	Brand accuracy check	Brand
16	Competitor benchmark	Competitive

Lesson 2.4: Reading and Interpreting an Audit Report

A professional GEO audit report always follows the same 5-part structure:

1. Executive Summary (1 page) — Overall score, grade, top 3 findings. For C-level.
2. Category Breakdown (2-3 pages) — Individual scores per category. For marketing managers.
3. Detailed Findings (3-5 pages) — Each check with status. For the tech team.
4. Quick Wins (1 page) — Immediately actionable measures.
5. Implementation Roadmap (1-2 pages) — Prioritized measures over 4-12 weeks.

Most common mistake: Beginners write the report for themselves, not for the recipient. A CTO doesn't need an explanation of what JSON-LD is. A marketing manager doesn't need code snippets.