Methodology

Everything on this site is designed to be checkable by a stranger who distrusts us. These are the rules we build to; the build fails when we break them.

Sourcing

• Primary first. Every timeline entry, milestone and recommendation links a primary source: legislation, a statutory inquiry or serious case review, Hansard, a court-documented outcome, or an official government, police or inspectorate publication. We link a secondary source only where no primary exists, and we say so.
• Link permanence. Where an official body's site has been retired (as with IICSA), we link the UK National Archives web archive rather than a mirror.
• Automated checking. Every source URL is machine-checked at build time. A small number of official domains (the parliament.uk family, nspcc.org.uk) block automated requests; those are verified manually and flagged in our checker's allowlist with the date last checked.

The evidence tiers

Every statistic carries one of four badges — explained in full here. The build scans every published page for numbers appearing outside a badged element and fails if it finds one. A badge of ESTABLISHED means a primary source published the figure, including its own stated uncertainty; it never means the figure is beyond revision.

Grading recommendations

• implemented — delivery is documented in the cited source (a review, inspection, statute or official publication).
• partial — accepted but incomplete, in progress, or blocked on change outside the owner's control. We use the owning reviewer's own words wherever one exists (for Telford, Tom Crowther KC's two-year review).
• ignored — rejected, or dropped without delivery.

Where published evidence does not track a recommendation individually, we grade conservatively and say the evidence is absent rather than inventing precision.

Grading milestones

Met requires documented delivery by the promised date. Missed means the date passed. Overdue counters count calendar days from the promised date. Promised dates come only from documents the owner published — terms of reference, statements to Parliament, formal responses.

Protecting victims and trials

• We never name victims beyond those who have publicly self-identified, and we do not publish graphic detail. Institutional failure can be established at pattern level.
• We do not cover live or active prosecutions. Concluded convictions and published reports only — nothing here should be capable of prejudicing a jury.

Independence

This site is independent, non-partisan and takes no funding from any political party, campaign or government body. The record documented here spans administrations of every party; we track the office, not the rosette.

Corrections

We correct in public. Errors, disputed gradings and new evidence go through the corrections page; material corrections are logged there permanently.