Field notes

How we sample onboarding files without drowning the evidence room

Practical sampling choices for audits for fintech onboarding files when volumes are high and high-risk corridors matter most.

Stacked manila folders with handwritten sampling tags on a compliance desk

When a remittance firm opens twenty thousand new customers in a quarter, an audit that promises to “review everything” is theatre. Blue Cedar Compliance Office builds sampling around corridors, product triggers, and known exception types so the evidence room stays usable.

Start with the risk map, not a percentage

We ask which corridors, customer types, and onboarding channels carried the most alerts last quarter. A stratified sample of forty files often reveals more than a random two hundred if those forty include agent-introduced customers, PEPs, and cases closed after a single screening hit.

Keep the worksheet boring

Each sampled file needs the same columns: screening timestamps, escalation notes, override approvals, and whether the customer was later subject to enhanced measures. Fancy spreadsheets invite narrative padding; a dull worksheet forces exceptions into view.

Close with owners, not adjectives

Findings should name the control owner and the missing artefact — “agent questionnaire absent for three of eight sampled introductions” — rather than calling the programme “immature.” Boards can fund a missing form; they cannot fund a vague adjective.

Back to field notes