GeoScore Engine Update: Multi-Page Scanning & CJK Support

Auditing AI readiness requires seeing your website the same way modern AI crawlers do. Today we deployed three major upgrades to our core scanning engine: multi-page structured data aggregation, direct HTTP security header inspection, and full CJK language accuracy.

These improvements eliminate false negatives on modern Single Page Applications (SPAs) and international websites, giving you a significantly more accurate audit score.

1. Multi-Page Discovery & Unified Schema Aggregation

Previously, scanning an origin only analyzed the specific root HTML payload. However, modern sites frequently distribute their JSON-LD schemas across specialized sub-routes:

• • /about: Hosts Person, Organization, and editorial byline credentials.
• • /review/ or /blog/: Contains Article and customer review schema.
• • /sitemap.xml: Verified with real status 200 HTTP checks.

The scanner now proactively checks linked trust routes and combines the metadata before computing your E-E-A-T and Structured Data sub-scores.

2. Real HTTP Security Header Inspection

Security is a foundational signal for search engine trust. Transport security directives like Strict-Transport-Security (HSTS) and Content-Security-Policy (CSP) are delivered as HTTP response headers rather than raw HTML strings.

Our fetch pipeline now captures the live HTTP response headers from origin servers and edge proxies, ensuring that valid server configurations receive full credit under the E-E-A-T security check.

3. Accurate Multi-Lingual & CJK Scoring

Non-Latin languages require customized tokenization and semantic heuristics. We have fine-tuned 8 analyzer behaviors for Chinese and CJK web properties:

• • CJK Word Count: Calculates information density with CJK character weighting (~0.6 equivalent English words per character) instead of relying solely on whitespace splitting.
• • Concise H1 Evaluation: Gives appropriate weight to compact, highly informative headings.
• • Localized Trust Signals: Accurately identifies author bylines (作者, 撰稿), trust pages (关于, 隐私, 协议), and testimonials (评价, 口碑, 反馈).
• • Localized Numerical Units: Recognizes stats cited with 万 and 亿 scales.