How the catalog works

Methodology

The catalog uses direct first-party public evidence for material provider facts. Discovery directories never become publication proof.

Evidence states

Disclosed means a current authoritative source directly states the fact. Provider claimed is visibly separate and never means issuer verified. Undisclosed renders as an em dash—not zero or false. Conflicting, expired, inaccessible, and previously-disclosed-but-missing facts remain preserved and are blocked or routed to review.

Provenance and freshness

Each published decision field resolves to a direct URL, exact quotation, retrieval date, HTTP status, snapshot reference, collector version, and SHA-256 content hash. Commercial and capability facts enter freshness queues; a changed hash creates a review candidate rather than silently changing meaning.

Dedupe and publication

Canonical domains and normalized names detect duplicate candidates. Merges require an audit event and preserve aliases and history. Profiles require current service evidence. Collections use one shared predicate: an approved single facet, at least eight current deduplicated facilities and four providers, current contextual evidence with allowed use, unique editorial context, and no unresolved conflict.

Ranking

Organic order uses evidence completeness, freshness, and fit. Payment, sponsorship, affiliate state, and claim ownership are not inputs. This directory does not designate “best” providers.

Research scope

The current package contains 40 unique Northeast-oriented profiles and 474 evidence rows, last verified August 3, 2026. Thirty-four profiles (85%) have at least three useful decision fields. Dynamic source pages can change; verify scope directly before making a commercial decision.