Sampling change tickets without drowning the team

Change control is where product velocity and audit evidence collide. Teams fear that sampling will freeze releases; auditors fear that “emergency” labels hide unreviewed production edits.

Agree the population first

Define the population as production-affecting changes in a named window — including hotfixes. Exclude purely documentary edits if your policy says so, and write that exclusion into the scoping memo.

Stratify, then sample

Separate high-impact payment or ledger changes from lower-impact content updates. Draw samples from each stratum. A random draw across an undifferentiated pile under-represents the risky class and over-tests the harmless one.

What “complete” looks like

For each sample, expect the request, approval, test evidence, and deploy record. Missing one piece does not always mean a high finding — but a pattern of missing approvals on ledger changes will. During fieldwork we share interim lists so teams can locate artefacts before the draft report hardens.