The template every case follows
- Context. What was audited and why.
- What we had published. The finding as delivered.
- What the site showed. The live observation, method named, frames scanned.
- The correction. What changed in the wording and what did not.
- What was measured after. Per consent state, decoded, counted two ways.
- What we got wrong. Our own retracted readings, kept on the record.
- Status. Build version, verification count, what remains unread.
- How it grades. The row the case would produce in an answer audit.
No client names. No vendor logos. Numbers only when measured. Where a number cannot be shared, the mechanism is.
Until the first case is cleared for publication, the method is on /how-we-grade, the measurement behind it on /tracking-audit and /cookie-audit, and the engagement on /answer-audit.