Operational guide ·
KYC workload: measure hours, not waiting days
An onboarding file can take 40 calendar days without consuming 40 days of staff work. Measure both, but keep them separate.
This guide helps private-bank compliance and operations teams estimate workload from their own figures. It provides no market Benchmark or compliance opinion.
1. Define the work being measured
Choose one period and one scope before collecting numbers. For onboarding, define when a file starts and when it is considered complete.
Calendar duration includes waiting for documents, client replies and internal decisions. Human working time adds the active effort of each contributor.
For example, 2 colleagues working together for 1 hour contribute 2 human working hours. A file waiting overnight adds elapsed time, not active effort.
Record document checks, follow-up, escalation and approval work if they belong to your chosen scope. Do not include waiting time as staff work.
2. Calculate each Area separately
Use matching units and periods. Annual volumes require hours per file; monthly alert volumes require minutes per alert, converted to hours.
Fictitious example · Not customer data or market reference figures
- Onboarding
- 300 files per year × 8 hours = 2 400 hours annually
- Lifecycle monitoring
- 500 alerts per month × 12 minutes ÷ 60 = 100 hours monthly
- Periodic review
- 800 reviews per year × 5 hours = 4 000 hours annually
These are workload estimates, not staffing requirements. They do not establish available capacity, productivity, regulatory adequacy or achievable savings.
Planned review volume estimates the work planned, not the work already completed. Keep those volumes separate when monitoring delivery.
3. Keep uncertainty visible
Mark each input as measured, estimated or unknown. Record the period, source and file types covered alongside the number.
For a measured average, record the number of files observed and how they were selected. A small convenience sample may miss complex cases.
Keep standard files and enhanced-due-diligence files separate when their effort differs materially. A blended average can hide the workload of complex cases.
If 300 files each require an estimated 6 to 10 hours, the annual workload range is 1 800 to 3 000 hours.
If volume or effort is unknown, leave the calculation unavailable. Unknown does not mean zero, and a precise-looking result does not improve an estimate.
4. Check overlap before adding totals
An alert can trigger a periodic review. If that review already includes alert analysis, adding both estimates can count the same activity twice.
Map shared activities and assign their time once. Until the boundaries are clear, present the 3 Area estimates separately rather than a grand total.
Likewise, count repeated document requests separately only when their effort is excluded from the original hours-per-file figure.
5. Turn the figures into a check
For onboarding, identify which missing documents cause repeat requests. For alerts, split effort into triage, investigation, escalation and recording closure evidence.
For periodic reviews, classify overdue files by risk and age against your own schedule. A workload estimate alone does not determine their priority.
Test one process change, then repeat the measurement using the same scope. Keep any tooling recommendation separate from the evidence of improvement.
Hypothetically, reducing 800 reviews from 5 to 4 hours releases 800 working hours annually at constant volume. This is not a promised Wecan saving.