The open ledger
Every revision to our methodology is versioned, dated, and explained here. When a factor's importance moves, you see exactly when and why — nothing changes in the dark.
- What changed
- Legitimacy & Accreditation becomes scored-with-disclosure: where third-party evidence exists it scores at full Critical weight, and where it does not the page states what we searched and that we could not establish it. The factor is NOT re-tiered — it remains Critical.
- Why
- A Critical factor that no obtainable evidence can satisfy is a specification error, not a standard. The evidence path was measured rather than assumed, and for a meaningful share of providers there is no third-party record that answers it. Telling a reader what we searched and could not establish is more informative than a page that cannot publish at all.
- Did scores move?
- Yes — live scores were recomputed under this version.
v5
Major- What changed
- 36-factor scoring config (consumer-sentiment de-scoped; universal 12→11), and the first version to freeze the rubric text alongside its weights.
- Why
- Closes the rubric-freeze gap: until now no MethodologyVersion referenced the rubric, so every rubric edit was retroactive across every published version — a reader could not see the standard a past score was judged against. v5 snapshots the rubric with the config, so the standard is frozen with the scores it produced.
- Status
- Published; superseded by v6 before any live score was computed under it.
v3
Major- What changed
- Operator-trust score. Additive to v2 — the factor spine, tiers, and weights are unchanged, so no vertical (per-treatment) score moves. It adds a new public score type: a brand-level operator-trust score on each provider hub, computed as the weighted mean of the company-level factors that measure the operator regardless of treatment — legitimacy and accreditation, clinical oversight and safety, transparency, consumer sentiment, policy fairness, data privacy, popularity, quality under growth, access speed, and support model.
- Why
- A provider hub needed one honest headline number that reflects the operator itself rather than any single treatment. Blending the vertical scores would double-count the company-level factors and break under sparse coverage; the operator-trust score instead reuses the shared-core signals already gathered, so the hub and each treatment scorecard never contradict each other on a brand-level factor. A new public number is a new claim — so it is a versioned, dated change.
- Did scores move?
- No — no live provider score changed under this version.
v2
Major- What changed
- Scoring calibration. Same factor spine, refined derivation: regulatory actions are now weighted by severity (marketing and labeling actions count far lighter than safety, quality, or licensing and fraud actions); complaint-skewed indices like the BBB are treated as weak context, not score drivers; third-party data we cannot retrieve is excluded and the score re-normalized over what was assessed, with a published completeness figure, so a provider is never penalized for our retrieval gap; and every adverse finding now requires a primary source.
- Why
- The launch calibration changed how a raw observation becomes a subscore while the factor set stayed fixed, so scores were already being computed under these rules — this version makes the public label honest. It also makes scores fairer (severity-aware, primary-source-backed) and robust to data we could not retrieve.
- Did scores move?
- No — no live provider score changed under this version.
v1
Major- What changed
- Initial ranking framework — the universal factor spine (importance tiers: Critical, High, Medium, Supporting) with per-category overlays. This was the lens the first scoring ran against; on its own it never produced a public score.
- Why
- Establish one open, never-for-sale standard so every provider is measured on the same independently-verifiable signals — not reputation, not ad spend.
- Did scores move?
- No — no live provider score changed under this version.