GEO Score Report: From 39 to 78 in 3 Weeks — A Complete Site Audit Retrospective

When a popular Chinese tech blog ran its first GEOScope audit, the result was sobering: 39 out of 100. Here's exactly what was broken, how each dimension was fixed, and what happened next.

Baseline: 39/100 — "AI Can't Understand This Site"

8 dimensions breakdown:

Crawler Access (2/12)

GPTBot and ClaudeBot blocked in robots.txt. The site was literally invisible to the largest AI engines.

Structured Data (3/14)

No JSON-LD on any page. AI engines had zero semantic context about page types, articles, or authors.

AI Discovery Files (0/10)

No llms.txt, no AI-TXT. A blank page from the AI's perspective.

Content Citability (8/16)

Articles were well-written but introduced slowly. AI engines scanning first paragraphs found no direct answers to extract.

Week 1 Fixes: Infrastructure

Updated robots.txt — allowed GPTBot, ClaudeBot, PerplexityBot, Google-Extended
Deployed Article JSON-LD to 50+ blog posts with author, datePublished, publisher
Created llms.txt with site description and top 20 article links

Score after Week 1: 52/100

Weeks 2-3 Fixes: Content

Restructured 10 highest-traffic articles: moved conclusions to first paragraph
Added FAQ sections to 3 cornerstone guides
Added author credentials to every post (name + bio + social link)
Linked to 2 academic references and 1 industry report per article

Score after Week 3: 78/100 — Up from Grade D to Grade A-

Key Lessons

The biggest single-impact fix is allowing AI crawlers — score jumped 13 points in one change
JSON-LD deployment is high-effort but essential for long-term AI comprehension
Content restructuring (conclusion-first) had the highest citability ROI

See your own GEO score in 30 seconds

Free audit, 8 dimensions, instant report.

Get Your GEO Score →