AI Search Audit (GEO Audit)
An on-demand audit of your website's readiness to be cited and recommended by AI.
Last updated: 2026-05-05
AI Search Audit (GEO Audit)
The AI Search Audit (formerly "GEO Audit") is a one-click crawl-and-grade of your website against the criteria that determine whether AI search engines (ChatGPT, Gemini, Perplexity, Google AI Overview) can find, trust, and cite you. Where AI Scans measure outcomes, the audit measures inputs — the technical and content readiness of your site itself.
What the audit covers
The audit is organised into three pillars:
1. Discoverability
Can AI crawlers actually reach and parse your site?
robots.txtrules (does it block AI user agents?)llms.txtpresence and qualitysitemap.xmlvalidity and coverage- HTTP status codes on key pages
- Render-blocking JavaScript
2. Authority
Do the signals AI models use to trust a domain check out?
- Schema.org structured data (Organization, Product, FAQ, HowTo)
- Open Graph and Twitter Card metadata
- Canonical tags
- Breadcrumb markup
- Author and publication metadata
3. Content Readiness
Is the actual content shaped the way AI models prefer?
- Heading hierarchy (single H1, semantic H2/H3)
- Paragraph length and answer-friendliness
- FAQ presence on key pages
- Internal linking depth
- Page-level summaries and tl;dr blocks
How it works
- You enter a domain (typically your own).
- Pranas samples a representative set of pages (homepage, top product/feature pages, blog index, a sample of articles).
- Each page is fetched twice — once via Google's user agent and once via Bing's — using Firecrawl, with a UA fallback if either is blocked. This catches the common case where a site serves different content to different crawlers.
- Each pillar is scored 0–100 with a per-check breakdown.
- The "Top Fixes" panel highlights the highest-impact issues.
Rate limit
Audits are limited to 1 per hour per account. The crawl is resource-intensive and the underlying scoring rarely changes within an hour. If you need to re-run sooner after a fix, wait the hour — the new score is more meaningful.
Pillar reference
| Pillar | Question it answers | Key checks | Weight in the composite |
|---|---|---|---|
| Discoverability | Can AI crawlers reach and parse the site? | robots.txt, llms.txt, sitemap.xml, HTTP status, render-blocking JavaScript | Highest |
| Authority | Do the trust signals check out? | Schema.org, Open Graph, canonical tags, breadcrumbs, author metadata | Medium |
| Content Readiness | Is the content shaped the way AI models prefer? | Heading hierarchy, answer-friendly paragraphs, FAQs, internal links, tl;dr blocks | Medium |
Each pillar is scored 0–100 independently with a per-check breakdown, and the composite is weighted rather than averaged.
Reading the score
Each pillar is independent. A site can score 90 on Authority (great schema) but 40 on Discoverability (broken robots.txt). The composite isn't an average — it weights Discoverability highest because if AI can't reach you, the other two don't matter.
Related articles
- Audit rate limit
- Top Fixes per pillar
- Which pages does the audit crawl?
- Citation optimization tactics