Eliciting Regulatory Reconciliation Knowledge Before Retirement

Sequence the work by risk and irreplaceability. First, shadow the retiring analyst through at least one complete reconciliation cycle and collect artifacts: spreadsheets, query logic, adjustment journals, exception logs, and email approvals. Use a critical decision method interview to surface cues, exceptions, and workarounds, not just the nominal process steps. Record the rationale for each adjustment and the source of truth for each figure. In parallel, issue a short pre-read to branches showing the outdated process map and asking them to mark differences before the workshops. Run the five branch sessions in 90-minute slots using a structured variant matrix: step, input, system, control, frequency, local exception, and regulatory impact. Do not cancel them. The retiring analyst knows the central reconciliation, but branch variants may explain audit differences that the central process assumes away. Vendor workshops should start after a draft baseline exists, otherwise the vendor will anchor requirements on incomplete central knowledge. The 200-page failure is a warning about format, not about elicitation: produce decision-focused outputs such as reconciliation rules, exception catalogues, and test cases with owners. Skipping branches would save perhaps two days but could leave undocumented variants that fail the four-month audit.

First classify each account: is it policy, a local workaround, a system limitation, or an error? Separate the regulatory requirement from local practice. Build a variant comparison showing frequency, transaction volume, control strength, audit evidence, and failure impact for each account. Use a facilitated session with the process owner, compliance, and branch representatives to decide the target standard; the decision right belongs with the process owner and compliance, not the project manager. Where variants are legitimate, document them in an exception register with conditions, compensating controls, and an expiry or review date. Test candidate rules against actual month-end reconciliations: reconstruct two or three historical periods and compare results. This exposes whether a seemingly minor variant changes balances. Do not average variants, because averaging controls can create a process nobody performs and no auditor can trace. Also capture rejected options and why they were rejected; that rationale prevents the vendor from reintroducing them later. The baseline should distinguish mandatory requirements, agreed local exceptions, and open decisions with owners and dates.

Validate by reconstruction, not by review alone. Ask the retiring analyst and a branch representative to walk through three historical reconciliations using only the draft rules and test cases. Any step that requires undocumented memory is a gap. Compare the reconstructed totals with legacy reports and investigate every difference. Package the knowledge as a small set of interlocking artifacts: a one-page process overview, a reconciliation rule table with source systems and owners, an exception catalogue, a control matrix mapped to audit assertions, and executable test cases with expected results. Each artifact should have a version, owner, status, and last-validated date. Keep narrative rationale where it explains a decision or a workaround, but avoid a 200-page prose document with no decision points. Hold a sign-off session with compliance and the vendor, record open items, and place the package in the project repository with controlled access and a change log. Schedule a pre-retirement handover check and a post-retirement contact arrangement if possible. The test is whether a new analyst can reproduce a month-end reconciliation and explain each adjustment without calling the retiree.