Structuring detection — are your thresholds too obvious?

Had an interesting conversation with our examiners last week. They challenged us on our structuring detection rules, specifically that we only flag transactions between $5,000 and $9,999. Their point: sophisticated launderers know about the $10k threshold and also know that banks watch the just-below-$10k range. So they structure at much lower amounts — $3,000, $4,000 — that fly under our radar.

Anyone detecting structuring money laundering at these lower amounts? How do you do it without drowning in false positives from normal small transactions?

Yeah, our examiners made the same point. We expanded our structuring detection to include any pattern of repeated cash transactions in round or near-round amounts regardless of the specific dollar range. Instead of a threshold-based rule, we use a velocity rule: X number of cash transactions within Y days that aggregate above $10k.

The false positive rate increased initially but after we added customer segmentation (excluding known cash businesses, filtering by expected activity level), it became manageable. We actually caught several structuring patterns at the $2,000-$3,000 level that we'd been missing for years.

The key insight: structuring money laundering patterns at lower amounts tend to involve more transactions per cycle, which actually makes them easier to detect statistically even though each individual transaction looks harmless.

Worth noting that some money laundering structuring has moved beyond cash entirely. We're seeing more structuring through prepaid card loads and P2P payment platforms where the reporting thresholds are different or less well-known to compliance teams. Make sure your detection isn't only looking at traditional cash activity.

Great discussion. One additional approach worth mentioning: peer group analysis. Instead of fixed thresholds, compare each customer's cash activity against similar customers. A restaurant depositing $8,000 three times a week looks normal compared to other restaurants. The same pattern from a consulting firm is an anomaly.

Peer group models catch structuring that threshold-based rules miss because the baseline is dynamic. The downside is they require clean customer segmentation data, which many institutions struggle with.

Another technique: round-amount detection. Structured deposits tend to cluster at round or near-round numbers ($4,900, $5,000, $7,500) more than legitimate deposits do. Tracking the distribution of deposit amounts by customer and flagging unusual roundness patterns can surface structuring activity at any dollar level.

The examiner pushback you described is becoming standard. Regulators are increasingly testing whether institutions can detect structured activity outside the $8,000-$9,999 band. If your rules only cover that range, you should expect a finding in your next exam.