Is Your Site AI-Ready? GEO & llms.txt Audit + Fix List

Audit your website for Generative Engine Optimization (GEO). Crawl your pages, test AI crawler access (GPTBot, ClaudeBot, PerplexityBot), validate llms.txt and Schema.org structured data, and score content answerability with Google Gemini 3.7 Flash.

---

🚀 Why Generative Engine Optimization (GEO)?

AI answer engines (ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews) are reshaping web search. Unlike traditional search engines that return 10 blue links, AI models extract direct facts, synthesize answers, and cite authoritative sources.

If your site:

• Blocks AI crawlers in robots.txt (e.g. GPTBot or ClaudeBot), AI models cannot read or update your content.
• Relies entirely on client-side JS (SPA shells), AI crawlers that do not execute JavaScript will see a blank page.
• Lacks structured data and Q&A formatting, AI answer engines will skip your pages in favor of competitors with clean schemas and direct answers.

GEO Site Auditor runs a comprehensive technical, structural, and semantic diagnostic across your website and generates an executive visual report (REPORT.htm) with ranked, actionable fixes.

---

🔍 What We Audit

1. Access & Crawlability (25% Weight)

• 13 Tracked AI Bots: Tests robots.txt access for GPTBot, ChatGPT-User, OAI-SearchBot, ClaudeBot, Claude-SearchBot, PerplexityBot, Google-Extended, Applebot-Extended, CCBot, Bytespider, meta-externalagent, cohere-ai, and Diffbot.
• HTTPS & Canonical Redirects: Ensures valid SSL certificate and clean non-looping HTTPS redirection.
• Sitemap Discovery: Detects XML sitemaps and sitemap indexes referenced in robots.txt or root paths.
• Custom 404 Response: Verifies that non-existent paths return HTTP 404 status rather than soft-404 false positives.

2. Machine Readability & Structure (30% Weight)

• llms.txt Standard: Checks for /llms.txt or /llms-full.txt (the emerging AI discovery standard).
• JavaScript Dependence & SPA Shells: Heuristically measures text-to-markup ratios and detects empty SPA containers to ensure static readability without headless browsers.
• Schema.org Structured Data: Validates JSON-LD and Microdata entities (Article, Organization, FAQPage, Product, TechArticle, etc.).
• Headings & Content Quality: Checks single <h1> presence, heading hierarchy gaps (e.g. H1 jumping directly to H3), word count depth, and image alt text coverage.

3. Trust, E-E-A-T & Freshness (15% Weight)

• Author Bylines & Person Schema: Scans for verifiable author signals and credentials.
• Content Freshness: Extracts and sorts <time datetime="..."> tags and JSON-LD datePublished / dateModified timestamps.
• Organization Identity: Validates official sameAs entity links (LinkedIn, X/Twitter, Wikipedia, Crunchbase).
• Essential Trust & Legal Pages: Multilingual keyword scanner for About, Contact, Privacy, and Terms pages across English, Balkan (BCS), German, French, Spanish, and Italian.

4. AI Content Answerability (30% Weight — Powered by Gemini 3.7 Flash)

• Zero-Hallucination Q&A Extraction: Extracts up to 3 high-impact user questions with exact verbatim quoted answer snippets from the source text.
• Direct-Answer First Detection: Evaluates whether your content answers user queries immediately or buries the answer in fluff.
• Topic Clarity & Scope: Scores how unambiguously the core topic is defined.
• AI Editorial Fixes: Concrete, actionable recommendations to improve citation probability in ChatGPT, Claude, and Perplexity.

---

📊 Transparent Pay-Per-Event (PPE) Pricing

You only pay for what you use. Failed or unrun checks are never charged.

Event	Price	Description
actor-start	$0.02	Base run initialization, robots.txt, sitemap parsing, and site-level checks
page-audited	$0.02	Per page crawled, analyzed, and evaluated with LLM answerability

Pages that fail to load are not charged, and platform usage is included — the prices above are all you pay.

Typical Run Costs:

• 5-page quick audit: ~$0.12 (12¢)
• 10-page standard audit: ~$0.22 (22¢)
• 25-page deep audit: ~$0.52 (~52¢)

---

📥 Input Parameters

Parameter	Type	Required	Default	Description
websiteUrl	String	Yes	—	Target website root URL or domain (e.g. https://example.com).
maxPages	Number	No	10	Maximum internal pages to crawl (1–50).
includeAnswerability	Boolean	No	true	Enable Gemini 3.7 Flash AI answerability scoring.
pageUrls	String List	No	[]	Explicit page URLs to audit (overrides crawler discovery).
geminiApiKey	String (Secret)	No	—	Optional BYOK Google AI Studio key to use your own quota.
geminiModel	String	No	gemini-3.7-flash	Override the LLM model ID used for answerability scoring.

---

📈 Visual Report & Output Formats

Every run outputs:

1. Visual Standalone Report (REPORT.htm): Open directly in Console under Storage → Key-Value Store → REPORT (Preview). Features an SVG score dial, priority fixes grid, collapsible page cards with cited Q&A excerpts, and print-ready PDF styling (Ctrl+P).
2. Apify Dataset: Structured tabular view of audited pages with HTTP status, dimension scores, and check details.
3. Key-Value Store: Raw SUMMARY.json for automated webhook integration.

---

🔗 The Complete AI Search Strategy

Once your technical foundation is audited, track your real-time brand citations across ChatGPT, Claude, Perplexity, and Gemini:

👉 Run AI Visibility Checker →

---

❓ Frequently Asked Questions (FAQ)

How is the overall GEO score calculated?

The GEO readiness score is a weighted composite (0–100) across Access (25%), Machine Readability (30%), LLM Answerability (30%), and Trust (15%). If LLM evaluation is disabled, the remaining three dimensions automatically renormalize to 100%.

What happens if my robots.txt blocks AI crawlers?

If your robots.txt blocks major AI bots (like GPTBot or ClaudeBot), the actor enforces a hard guardrail cap: the Access dimension is capped at 35% and the overall site score cannot exceed 48/100 ("Needs Work"). A site that AI models cannot crawl cannot receive a passing GEO score.

Is my BYOK Gemini API key safe?

Yes. API keys are encrypted at rest by the Apify platform, passed securely to the actor environment during runtime, and used solely to query the Google Generative Language API. Keys are never printed in logs or stored in dataset results.

Does the crawler execute JavaScript?

No. Most AI search crawlers (like GPTBot and ClaudeBot) crawl static HTML to conserve resources. GEO Site Auditor inspects raw server-rendered HTML using heuristic SPA detection to ensure your site is readable without heavy client-side JavaScript rendering.