Docs · Getting started
Understanding geoScore
A composite 0–100 metric we compute on every page. 5 inputs, weighted by AI-citation-lift correlation we measured across 4,000+ pages on bacwaterdepot.com.
JSON-LD schema coverage
How it's calculated: Award points for FAQPage (12), HowTo (10), Article (5), Product (3) JSON-LD blocks detected on the page.
How to lift it: Inject the missing schemas. FAQPage + HowTo together = +22 points for one PR.
Direct answer in first 200 words
How it's calculated: Regex-match for buyer-intent answer signals: 'yes', 'no', 'the answer is', 'standard ratio', 'typically', 'USP', 'FDA', and unit-prefixed numbers (e.g. '2 mL', '5 mg').
How to lift it: Rewrite the opening paragraph to lead with the literal answer + numbers. No throat-clearing.
Table density
How it's calculated: Award 8 points for one <table>, 4 more for two, 3 more for three+. Tables are the format AI engines extract most reliably.
How to lift it: Add a side-by-side comparison or specification table to each page.
Word count sweet spot
How it's calculated: Sweet spot 800–2500 words = full 15. 500–800 = 10. 2500–4000 = 12. 300–500 = 5. <300 or unbounded = 0.
How to lift it: Expand thin pages to 800+ with structured detail, not filler.
Internal link health
How it's calculated: 5+ internal links = 6 points. 10+ = 4 more. Signals hub-ness to crawlers.
How to lift it: Add a 'Related' section linking to your nearest topical neighbors.
H1/title format
How it's calculated: Question-format H1 or title (contains '?') = 5 points. Contains buyer-intent words (how, what, where, when, why, which, best, vs, compare) = 5 more.
How to lift it: Reframe statement-style titles into the literal buyer question.
Threshold guidance
- · 0–30: page is invisible to AI engines. Schema + content-format work needed.
- · 30–60: below AEO threshold. Pages get crawled but rarely cited.
- · 60–80: citable. Should appear in some AI-engine answers within 30 days.
- · 80+: high-priority citation candidate. Often top-3 cited source for buyer-intent queries.
