← Guides

Guide

What "verified" means on scrapebench

How the nightly golden-diff canary checks each government-data actor against its live source portal.

Every actor listed here scrapes a live government portal — a state licensing board, a federal opportunities feed, an OSHA inspection record. Those sources change without notice: a column gets renamed, a form field moves, a portal is taken down for maintenance.

The nightly check

A canary runs each published actor against its real source every night and diffs the result against a frozen “golden” fixture. If the shape still matches, the run is a pass. If the source drifted, it’s a fail — and a self-healer opens a fix.

The success percentage on the health board is the share of those nightly runs that passed. “Insufficient data” means there haven’t been enough runs yet to compute a rate — it never means 0%.

What it does not claim

This is a correctness signal, not a traction or performance claim. It does not report run timings, record counts, or revenue — those aren’t measured here, so they aren’t shown.

Compare exactly which fields each actor returns on the coverage matrix.