Skip to main content

Lesson 16 of 25

Auditing CDD, EDD, and KYC File Reviews

4 min read · CAMS-Audit

Run risk-based file reviews that surface missing, stale, and contradictory data, test whether EDD is performed and not just promised, and re-rate customers to expose flawed models and unjustified overrides.

CIP, CDD, EDD — the ladder

  • CIP — identify and verify who the customer is
  • CDD — understand the customer and their expected activity
  • EDD — extra scrutiny for higher-risk customers
  • Audit tests each rung against risk

Customer due diligence is a ladder, and the exam wants you to know which rung applies when. At the bottom is the Customer Identification Program, or CIP, under thirty-one CFR ten-twenty point two-two-zero: identify and verify who the customer is. Above that is customer due diligence proper, understanding the nature and purpose of the relationship and the customer's expected activity so you can spot deviations.

At the top is enhanced due diligence, EDD, the extra scrutiny applied to higher-risk customers like politically exposed persons, correspondent banks, and high-risk geographies. FATF Recommendation ten frames CDD this way, and audit's job is to test that each rung is applied to the right customers and actually performed.

The risk-based file review

  • Sample by risk — over-weight high-risk customers
  • Check identity verification and beneficial ownership
  • Confirm risk rating matches the file's facts
  • Look for missing, stale, or contradictory data

The core fieldwork technique here is the file review, and it should be risk-based. Rather than a flat random sample, over-weight high-risk customers, where the consequences of a gap are greatest, while still sampling lower-risk files for baseline assurance. For each file, check that identity was verified, that beneficial ownership was collected and verified for legal entities, and that the expected-activity profile makes sense.

Then confirm the customer's risk rating actually matches the facts in the file. And hunt for the three classic defects: missing data, stale data that was never refreshed, and contradictory data, where the file says one thing and the system says another. Each is a finding, and patterns across files point to a process problem.

Auditing EDD where it matters most

  • Is EDD triggered for the right customers?
  • PEPs, correspondents, high-risk geographies, high cash
  • Source of wealth and source of funds documented?
  • EDD on paper vs. EDD actually performed

Enhanced due diligence is where the highest-risk customers live, so test it hard. First, is EDD actually triggered for the right customers, politically exposed persons, correspondent banking relationships, high-risk geographies, cash-intensive businesses? A common failure is EDD criteria that exist on paper but never fire in the system.

Second, where EDD applies, is the deeper information really gathered, source of wealth and source of funds documented and reasonable, ownership structures unwound, ongoing monitoring intensified? The gap to watch for is EDD on paper versus EDD performed: a policy that promises enhanced scrutiny for PEPs, paired with PEP files that hold nothing more than a standard customer's. That gap is exactly what the exam likes to surface.

Auditing risk rating and overrides

  • Is the rating methodology applied consistently?
  • Do manual overrides have documented justification?
  • Watch for ratings downgraded without basis
  • Re-rate a sample yourself and compare

Finally, audit the customer risk-rating engine and its overrides. Check that the rating methodology is applied consistently, so similar customers get similar ratings, and that the factors feeding it are accurate. Then scrutinize manual overrides, where a human changes the system-generated rating.

Overrides aren't inherently wrong, but each one needs a documented, reasonable justification. The red flag is a pattern of customers quietly downgraded from high to medium risk with no basis, which conveniently reduces monitoring and EDD obligations. A strong technique is to independently re-rate a sample of customers yourself and compare your result to the system's; systematic gaps reveal either a flawed model or improper overrides.

Ongoing CDD and trigger events

  • CDD isn't one-and-done — profiles must stay current
  • Trigger events: ownership change, activity shift, adverse news
  • Periodic review cadence tied to risk level
  • Test whether updates actually happen and re-rate the customer

A point the exam emphasizes is that customer due diligence is not a one-and-done event at onboarding; the CDD Rule requires ongoing monitoring to keep customer information current. So audit tests the ongoing side, which is where many programs quietly fail. Are trigger events handled, a change in beneficial ownership, a sharp shift in transaction activity, adverse media or a sanctions hit, all of which should prompt a profile refresh?

Is there a periodic review cadence tied to risk, with high-risk customers reviewed more often than low-risk ones? And critically, do those updates actually happen on schedule, or has a backlog of overdue reviews built up while customers' real risk changed beneath stale profiles? When you find a high-risk customer whose file hasn't been reviewed in years despite changed activity, that's a finding, and a stale profile that never fed back into the risk rating means the whole monitoring posture for that customer is built on outdated facts.

Ongoing CDD is where onboarding diligence either stays alive or silently decays.

Recap and next

  • CIP, CDD, EDD — apply the right rung by risk
  • Risk-based file reviews surface missing/stale data
  • Test that EDD is performed, not just promised
  • Next — auditing transaction-monitoring systems

Recapping: customer due diligence is a ladder from CIP to CDD to EDD, and audit tests that the right rung is applied to the right customer and genuinely performed. Risk-based file reviews surface missing, stale, and contradictory data, EDD testing exposes the gap between enhanced scrutiny promised and performed, and re-rating a sample reveals flawed models or unjustified overrides. Next, we move to one of the most technical fieldwork areas on the exam: auditing transaction-monitoring systems, including coverage and tuning.

Take the CDD practice questions first.

Sources

  • FinCEN CDD Rule, 31 CFR 1010.230 — beneficial ownership
  • FFIEC BSA/AML Examination Manual — Customer Due Diligence and risk rating
  • FATF Recommendation 10 — customer due diligence
  • 31 CFR 1020.220 — CIP

Test your knowledge

A few CAMS-Audit questions on this material — pick an answer to see the explanation.

  1. Q1. An engagement planning memo defines the scope as all CDD processes at three named branches over the 12 months ending December 31. During fieldwork, two additional branches are identified with identical CDD processes. Should the auditor expand scope?

  2. Q2. To avoid duplicating compliance's QA work, an auditor decides to rely entirely on the second line's CDD file review results as her own testing evidence. What is the flaw?

  3. Q3. An auditor samples 50 customer risk-rating files using stratified random sampling: 20 from the high-risk stratum and 30 from the medium-risk stratum. She finds zero exceptions. Which conclusion is defensible?

  4. Q4. An auditor finds a single exception in a statistical sample of SAR narratives — one narrative is a copy-paste of another with only the dates changed. How should she treat this exception?

Ready to practice?

Put this lesson to work on real CAMS-Audit questions.

Drill the full CAMS-Audit bank →