How we know what we claim
Provenance is not fine print here; it is the point. Every published claim traces to a content-hashed snapshot of a public document, stored when the claim was made, so the evidence cannot quietly change under the reader.
No source, no claim
When a source is ingested, Fundprint fetches and stores the exact document as a content-hashed snapshot, then links that snapshot to the claim it supports. A row missing its source records does not publish, regardless of how confident the match is.
Historical ownership that has ended is not shown as current. It is recorded as a dated event with its own source and placed on a timeline, not rendered as a live ownership link.
Each snapshot is fingerprinted. The same URL captured twice with the same content is stored once; a change produces a new record.
The four kinds of source
- NPPES provider registry
The federal NPI registry. The primary evidence that a clinic exists as a provider organization.
- Owner location directories
An owner's own public directory of centers, read from machine-readable structured data. The supplement that fills NPPES's undercount.
- Acquisition announcements
Private-equity firm and company statements that assert an ownership or a deal, captured at the source.
- Reputable trade press
Named reporting used to corroborate an ownership link where a primary statement is thin.
One claim, its evidence
The full rules live in the versioned methodology. How the machine works