Case Studies and Writing Samples
Case rounds · writing samples · grading criteria · last reviewed September 2, 2026
Most financial crime hiring decisions turn on one exercise: a case put in front of you, and a short window to say what you would do about it. The panel is not looking for the right answer — they are looking for a process that would survive a quality review, an examiner, and, in the worst case, a subpoena.
This page reorganises the worked case exercise and writing-sample guidance from the education and training pillar into interview-prep order: how the round runs, what the grader marks, how to write under time pressure, and where candidates lose the seat.
How case rounds are run
Case rounds arrive in three common shapes. The live walk-through is the most frequent: an interviewer reads out an alert or a short fact pattern and asks you to think aloud. There is no document to hide behind, and the pace is conversational — expect interruptions, and treat each one as a prompt rather than a correction.
The take-home is the second shape: a one- or two-page fact pattern with a deadline measured in hours or a couple of days, returned as a written disposition. Take-homes are graded harder than live rounds because you had time; a loose structure or an unsupported conclusion reads worse when the clock was generous.
The timed on-site exercise is the third: thirty to sixty minutes in a room or a shared document, sometimes with system screenshots, producing a written recommendation. This is the closest analogue to the job, and the one most candidates prepare least for.
In all three, the fact pattern is usually a composite rather than a real file — deliberately incomplete, with one or two details that cut against the obvious reading. The gaps are the exercise. Naming what you do not know is scored; inventing it is not.
What graders look for
Panels rarely publish a rubric, but the marking behaviour is consistent across institutions. Five things get scored.
Frame before conclusion
A strong answer names what the activity is before saying what should happen to it. Deposits under the reporting threshold across multiple branches, followed by rapid movement out, is a structuring-plus-flow-through pattern; say that first, then reason from it. Candidates who open with "I'd file" have skipped the part being graded.
Evidence you would gather
Name the specific checks: customer profile and stated occupation, account opening documentation, prior alert and case history, counterparty relationship, negative news, and the branch or channel pattern. A list that could apply to any case reads as memorised; a list tied to the fact pattern reads as practice.
Disconfirming evidence
The single most reliable separator. Say plainly what would change your mind — a cash-intensive business with a documented deposit history, an expected third-party payment, a life event that explains the timing. Candidates who cannot construct the innocent explanation are the ones who over-file, and every panel knows it.
A decision with its basis
Reach one. Escalate or close, and state the basis in a sentence a reviewer could adopt. Hedging past the end of the exercise reads as an inability to own a file — and owning files is the job.
Confidentiality discipline
Graders notice how you talk about filings. Discuss the typology, the decision logic, and the escalation path; never disclose that a specific report was or was not filed on a real matter, and never bring real case facts into the room. Restraint here is scored as professional maturity, and its absence is disqualifying on its own.
Structuring a written narrative under time pressure
Written rounds reward a fixed skeleton, because the skeleton lets you spend your thinking on the facts instead of the format. Use the same order every time and the writing takes care of itself.
The five-block skeleton
011. Subject and relationship
What they're testingWho the customer is, when the relationship began, what the account is used for on its face, and what the stated profile says the activity should look like.
022. Activity
What they're testingWhat happened, in dates and amounts, in chronological order. No adjectives. A reviewer should be able to reconstruct the pattern from this block alone.
033. Why it is unusual
What they're testingThe gap between the stated profile and the observed activity, named as a typology. This is the analytical core and the block most candidates rush.
044. What was checked and what remains open
What they're testingThe verification you performed, the explanation you tested, and the questions you could not close. Unresolved gaps belong in the file, not out of it.
055. Conclusion and next steps
What they're testingThe recommendation, its basis, and what happens next — escalation, continued monitoring, a documented close, or a relationship review referral.
Time allocation on a sixty-minute exercise
Roughly ten minutes reading and marking the fact pattern, ten building a rough timeline of dates and amounts, twenty-five drafting straight through the skeleton without editing, and the last ten reading only for two things: unsupported assertions and sentences a reviewer could misread. Do not restructure in the final pass — a complete plain draft beats an elegant half-finished one every time.
Sentence-level habits that read as experienced
Write dates and amounts, not "several" and "large". Attribute every fact to where it came from — the system, the document, the interview. Use plain declaratives; conditional pile-ups signal uncertainty you do not intend. And write for a reader two years from now who has none of your context, because in practice that is exactly who reads it.
Common failure modes
Conclusion first
Announcing an outcome in the opening sentence and reverse-engineering support for it. Graders read the rest as advocacy rather than analysis.
Typology name-dropping
Labelling the pattern with a term — layering, smurfing, trade-based — without showing the facts that fit the label. The vocabulary is cheap; the mapping is the skill.
Boiling the ocean
A twenty-item checklist of everything that could theoretically be reviewed. In a real queue that is a backlog. Prioritise, and say why those checks first.
Silence on the innocent explanation
Never testing the benign reading. It suggests you would escalate everything, which is a quality-assurance problem and a resourcing problem at once.
Confidentiality slips
Referencing real filings, naming real customers, or bringing sample work that was never redacted. This ends candidacies regardless of the rest of the answer.
No written artefact
Arriving without a sample when the panel asks for one. Almost every candidate is in this position, which is exactly why the ones who are not stand out.
Building your writing sample
Prepare one before you need it: a single page showing a decision, its evidentiary basis, and its escalation path, written in the five-block skeleton above. It answers the question every hiring manager actually has — can this person write a file that survives review.
If everything you have written is confidential, do not redact your way toward a sample; build a clean one from a public source instead. Take a published enforcement action or an indictment, reconstruct the pattern as if it had surfaced as an alert on your desk, and write the disposition. Summarising a primary document cleanly is itself the demonstration, and the provenance question disappears.
Bring it printed, keep it to one page, and be ready to say in one sentence what decision it records. A sample you cannot summarise verbally works against you.
Prepare further
Frequently asked
What is a financial crime case study interview?
A fact pattern — an alert, a transaction sequence, or a customer file — that you are asked to analyse and dispose of, either aloud, as a take-home, or as a timed written exercise. Panels grade the process: how you frame the activity, what you would check, what would change your mind, and whether your decision carries a stated basis.
How do I structure a SAR-style narrative under time pressure?
Use a fixed five-block skeleton: subject and relationship, activity in dates and amounts, why it is unusual, what was checked and what remains open, then conclusion and next steps. The skeleton frees your thinking for the facts, and reviewers can follow it without your context.
Can I use real work as an interview writing sample?
No. Case facts, customer details, and anything filing-adjacent stay inside the institution — including in redacted form. Build the sample from a public enforcement action or indictment instead; reconstructing one cleanly demonstrates the same skill with none of the exposure.
What makes candidates fail case rounds?
Leading with the conclusion, naming a typology without mapping the facts to it, listing every conceivable check instead of prioritising, never testing the innocent explanation, and any slip on confidentiality. Each is avoidable with one rehearsed structure.