The A-F scoring spec, in detail.
Five standards every supplement or comparison post must clear before a tier is published. The rubric is versioned like software; this is the long-form spec, aligned to lib/content/evidence-tiers.ts.
- 01Standard 1
Grade A requires replicated RCTs
An A demands multiple high-quality randomised controlled trials with a consistent effect, replication in real-world settings, and a well-supported mechanism. Creatine monohydrate qualifies on 300+ RCTs (Kreider et al. 2017). A B is a single high-quality RCT, or several smaller RCTs pointing the same way, with a plausible mechanism.
- 02Standard 2
C, D, and F are graded down by evidence quality
C: mechanism plausible, observational data supportive, RCT evidence mixed or limited. D: mechanism speculative, observational evidence weak, RCTs neutral or negative. F: marketed claims contradicted by published data, mechanism implausible. The grade tracks the evidence, not the popularity of the molecule.
- 03Standard 3
Cited to primary sources
Each tier ships with a written rationale and a list of primary-source URLs (PubMed, ClinicalTrials.gov, NEJM). Every number, the dose, the n, the effect size, is sourced. A tier with no citation list does not publish.
- 04Standard 4
Priced in $/g and re-checked quarterly
Each reviewed product carries a $/g (or $/mg of bioactive) figure benchmarked against Costco, Amazon, and DTC on a dated check. Prices are re-checked quarterly; a move over 15% triggers a same-week update and the old figure is annotated, not overwritten.
- 05Standard 5
Independent, and corrected in the open
Tiers are set before affiliate availability is checked and are never invoiced. A tier reverses on a contradicting Certificate of Analysis, an independent lab failure, or a new meta-analysis. When that happens the post carries a dated note and the prior tier stays in the change log.
- v1.2 · April 2026 · $/g price math standardised on Costco / Amazon / DTC benchmarks with mandatory as-of date. 15% quarterly drift threshold added. lastReviewed date now rendered on every tier.
- v1.1 · February 2026 · 'What would change our mind' criteria documented (COA contradiction, independent lab failure, new meta-analysis). Registered-dietitian sign-off introduced.
- v1.0 · January 2026 · Initial public spec, A-F tier definitions, primary-source citation requirement, independence rule (rankings decided before affiliate checks).