What’s the 0-100 score formula?

Three weighted components:

• Required-property completeness (60% of score). Every required property present and correctly typed = full 60 points. Each missing required property subtracts ~15 points (depending on which property; image is heavier than alternateName). A datatype mismatch on a required property scores the same as missing.
• Recommended-property completeness (30% of score). Each recommended property present = +N points where N is proportional to its rich-result value. Adding aggregateRating on Product is worth ~6 points; adding color is worth ~1 point. The top 5 recommended properties together typically account for ~80% of the 30-point pool.
• Warning-count penalty (10% of score). Up to -10 points based on warning count (deprecated properties, non-canonical @context, suspicious values). Each warning subtracts ~1 point capped at -10.

Score bands:

• ≥85, green ring. Rich-result eligible. Cross-check with Google’s tester before shipping.
• 60-84, amber ring. Schema is valid but rich-result eligibility uncertain. Likely missing high-value recommended properties.
• <60, red ring. Errors present; rich result will not appear until errors are fixed.

The score is type-specific, a Product score of 92 means “ready for Merchant Listings + star snippets”; an Article score of 92 means “ready for Article + Top Stories”.