Grail Score measures platform quality, not expected financial return. A higher signal does not mean users should expect to profit.
What Grail Score Measures
Grail Score covers six dimensions: Fairness & Transparency (20%), Value & Liquidity (25%), Trust & Reliability (25%), Prize Quality (15%), Consumer Protection & Access (10%), and Product Experience (5%). As of version 1.2, each dimension is shown as its own icon badge with a plain-language state — Strong Evidence, Mixed Evidence, Limited Evidence, or Not Yet Assessed — rather than being blended into a single composite number. We made this change because folding six very different kinds of evidence into one digit hid more than it revealed: a thin evidence base on one dimension could make an otherwise well-evidenced operator look worse overall, and vice versa. Each dimension's state comes only from that dimension's own scored criteria.
What Grail Score Does Not Measure
Grail Score does not estimate expected value, odds of winning, or likely financial return — those are separate, offer-level figures (where published) shown on individual box pages, never folded into Grail Score itself.
Affiliate relationships, advertising relationships, sponsorships and popularity never affect Grail Score. An operator cannot pay to raise its score, and Grail Turf's commercial relationships are managed entirely separately from scoring.
Evidence Coverage
Every score displays an Evidence Coverage percentage — how much of that operator's score is currently backed by independently verified evidence versus operator-reported or unverified information. A low or 0% evidence coverage means the score is illustrative and should not be treated as a verified rating. We show this number honestly rather than hiding it.
Confidence Levels
Underlying facts that feed into Grail Score (buyback terms, published odds, shipping regions, and similar) are tracked with a confidence level: verified (independently confirmed against a primary source), operator-reported (stated by the operator, not independently checked), credible-secondary (reported by a reputable third party), or unverified. We do not upgrade a fact's confidence level without real evidence, and we do not fabricate a source to make a fact look more credible than it is.
Version and Review
This is Grail Score V1. Methodology, weights and scoring will evolve as Grail Turf's evidence-gathering process (Pull Scout) matures — any material change to weights or methodology will be reflected in a new score version, and prior scores will be labeled with the version they were produced under.
Version 1.1 (2026-10-15) is an interpretive clarification, not a change to weights or the aggregation rule: it documents how five specific product characteristics are scored consistently. Settlement in a cryptocurrency (e.g. USDC) is judged on disclosure, fees, usability and redemption reliability, never docked simply for not being a bank transfer. A fixed, non-negotiable buyback rate is judged on its disclosed basis, costs and whether it's honored, not treated as a weakness by itself. Support is scored on confirmed channel availability separately from responsiveness — a single negative review is a reported limitation, not proof a channel doesn't work. Independent reputation signals (e.g. review platforms) are read holistically — volume, recency, recurring patterns, operator responses — never mechanically rescaled from a star average. A criterion with no evidence is excluded from its dimension's average; it is never scored as a failure, and never scored as a pass either.
Version 1.2 (2026-10-15) is a presentation change: the overall numeric composite is retired in favor of six independent dimension badges, each in its own plain-language state. Weights, criteria, and how evidence is judged are unchanged from 1.1 — only how the result is shown to visitors changed.