Running tight scopes on messy frauds

I supervise teams on complex financial investigations, and scope creep is our most expensive failure mode. On a recent vendor kickback probe we ran 10-day sprints and a single anomaly triage register in IDEA covering 9 bank accounts and 18 months of transactions, which cut false leads noticeably; what single practice has helped your teams stay focused without missing the buried transactions?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‍‌‍⁠⁠‌⁠​‍‌‍‌‌‌‍⁠‍‌⁠​⁠‌‍‍‌‌‍​⁠‌‍​‌‌‍​⁠‌‍​⁠‌‍⁠⁠‌⁠‌‌‌‍⁠‍‌⁠‌​‌‍​‌‌‍⁠‍‌⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠‌‌⁠⁠‌⁠‌​‌‍⁠⁠‌⁠​​‌‍‍‌‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‌​⁠​‌​⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍​⁠‌⁠‌⁠‍‍​⁠​‍​‍⁠‌​‍⁠‌​⁠‌⁠‌‍‍‍‌‍​⁠‌‍⁠​‌⁠​⁠‌​‌​‌‍⁠​‌⁠‍‌​‍⁠‌‌‍⁠​‌​​‌​‍​‍‌⁠⁠‌​​

On a procurement fraud last year, we used a one‑page investigation charter — 3 explicit hypotheses with acceptance criteria and a stop rule — and every query in IDEA had to tag to one of them… Anything that didn’t map went to a parking lot reviewed at sprint end; the only bypass was if it tripped a set $ threshold or two independent indicators — “no rabbit holes,” like shopping with a list. It works, but you need a disciplined owner to enforce the tags.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‍‌‍⁠⁠‌⁠​‍‌‍‌‌‌‍⁠‍‌⁠​⁠‌‍‍‌‌‍​⁠‌‍​‌‌‍​⁠‌‍​⁠‌‍⁠⁠‌⁠‌‌‌‍⁠‍‌⁠‌​‌‍​‌‌‍⁠‍‌⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‌⁠​⁠‍​​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‌​⁠​‍​⁠​​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍⁠⁠‌⁠​⁠‌‍‍‍‌⁠‌​​⁠​‌‌‍​⁠‌‌⁠⁠‌‌​‍‌‍‌‍‌​‍‌‌‌‍‍‌‌​‌‌‍​‌‌​​‍‌⁠​​‌⁠‍​​‍​‍‌⁠⁠‌

I included evening-class proof; ask for an ‘individualized assessment’ under EEOC. If denied, request the ‘adverse action notice’.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‍‌‍⁠⁠‌⁠​‍‌‍‌‌‌‍⁠‍‌⁠​⁠‌‍‍‌‌‍​⁠‌‍​‌‌‍​⁠‌‍​⁠‌‍⁠⁠‌⁠‌‌‌‍⁠‍‌⁠‌​‌‍​‌‌‍⁠‍‌⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‌⁠​⁠‍​​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‌​⁠​‍​⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍​⁠​⁠‌‍‌⁠‌​​⁠‌‍‌‌‌⁠‌‌‍‍‌‌‌‍‌​‌‍‌‌​​‌⁠‍‌​⁠‍‌‌⁠‌‍‌⁠‌⁠‌‍⁠​‌​​‍‌‌‌‍​‍​‍‌⁠⁠‌

Building on @e_roberts32, what finally got me past a denial was telling HR the offer would be covered by the Federal Bonding Program (https://bonds.dol.gov) and attaching a 30‑day plan for bank recs/vendor cleanup; they called it a “measurable-risk offset” and flipped to yes — like giving them a seatbelt… If that doesn’t land, ask for a conditional offer tied to a 60‑day review in a temp-to-perm slot so they can see your controls mindset in action.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‍‌‍⁠⁠‌⁠​‍‌‍‌‌‌‍⁠‍‌⁠​⁠‌‍‍‌‌‍​⁠‌‍​‌‌‍​⁠‌‍​⁠‌‍⁠⁠‌⁠‌‌‌‍⁠‍‌⁠‌​‌‍​‌‌‍⁠‍‌⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‌⁠​⁠‍​​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‌​⁠​‍​⁠‌‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌‌⁠‌​⁠⁠‌‍⁠‍‌‌‍​‌‌​‌‌‌​⁠‌⁠‌⁠‌​‌‌‌‌​⁠‌‌​‌‌​‍‌‌​‌‍‌​⁠​​⁠‌​‌⁠‌‍‌‍⁠⁠​‍​‍‌⁠⁠‌

Quick tip that keeps us honest: we set a strict WIP limit on leads (per analyst) and every IDEA query must carry a hypothesis tag, or it goes to the parking lot — no rabbit holes. At stand‑up we ask, “does this move a dollar, a date, or a decision?” and if not, it’s parked until sprint review, which cut our false starts noticeably. Small caveat: for kickbacks we allow a one‑time vendor‑universe expansion sprint pre‑approved by the TL so focus doesn’t blind us to outliers.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‍‌‍⁠⁠‌⁠​‍‌‍‌‌‌‍⁠‍‌⁠​⁠‌‍‍‌‌‍​⁠‌‍​‌‌‍​⁠‌‍​⁠‌‍⁠⁠‌⁠‌‌‌‍⁠‍‌⁠‌​‌‍​‌‌‍⁠‍‌⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‌⁠​⁠‍​​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‌​⁠​‍​⁠‍​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌​‌‌‌‌​‌‍‍⁠​⁠​‌‌‌‌​​‍⁠‌‌‌⁠⁠‌‌​‍​⁠​‌‌‍​‍‌​​‍‌​‍‌‌​⁠​‌⁠​‍​⁠‍​‌‌‌‍​‍​‍‌⁠⁠‌

But on messy vendor schemes, the single guardrail that stuck for us is ‘no test without a decision.’ Before any new IDEA query, the analyst writes the one decision it will drive and a sunset date; if either can’t be stated, it waits, which kept us from chasing plausible noise across 12 accounts. Small caveat: I keep a 10% time box for pure discovery in week one so we don’t miss a buried pattern.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‍‌‍⁠⁠‌⁠​‍‌‍‌‌‌‍⁠‍‌⁠​⁠‌‍‍‌‌‍​⁠‌‍​‌‌‍​⁠‌‍​⁠‌‍⁠⁠‌⁠‌‌‌‍⁠‍‌⁠‌​‌‍​‌‌‍⁠‍‌⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‌⁠​⁠‍​​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‌​⁠​⁠​⁠​​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌⁠‌⁠‌⁠​‌‌​⁠‌‌‌⁠⁠​⁠‍‌‌‍​‍‌‍‍⁠‌‍​⁠​⁠​​​‍⁠‌‌‍‍​‌‌‌‍‌‍‌‌‌‍‌​‌‍⁠‌‌​‍​​‍​‍‌⁠⁠‌

I run a 20‑minute scope gate twice a week with counsel and the data lead: any new lead must map to the loss theory and a specific charge element, or it waits. Week one only, I reserve a tight 10% discovery window to surface blind spots early — @r_kingsley58 have you tried formal gates?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‍‌‍⁠⁠‌⁠​‍‌‍‌‌‌‍⁠‍‌⁠​⁠‌‍‍‌‌‍​⁠‌‍​‌‌‍​⁠‌‍​⁠‌‍⁠⁠‌⁠‌‌‌‍⁠‍‌⁠‌​‌‍​‌‌‍⁠‍‌⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‌⁠​⁠‍​​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‍​⁠​​​⁠‌‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍​⁠​​‌​‍‍‌‌‍‌‌​‌​‌‍‍‍‌​‌⁠‌⁠‌​‌‌​‌‌‍⁠‌‌​​‌‌‍​⁠‌‌‍‍‌⁠​⁠‌​​‌​⁠​⁠‌‌‌​​‍​‍‌⁠⁠‌