Six weeks to nine days: re-engineering a quarterly close

Profile: multi-family office servicing several families across 40+ entities. Details are anonymised; the workflow is representative.

01The situation

The quarterly close took six weeks. By the time reporting reached families, the numbers were already stale, and the team spent the gap between quarters recovering rather than analysing. The close was almost entirely manual: data pulled by hand from custodians and banks, pasted into linked spreadsheets, reconciled line by line, with each family's package assembled bespoke at the end.

02Where the six weeks went

Mapping the process honestly was the first step. The bulk of the time was not analysis — it was data movement and error-chasing. Feeds were re-keyed. Breaks were discovered late and investigated under time pressure. Every entity's eccentricities lived in one person's head, so nothing could be parallelised and nothing could be delegated safely.

03The re-engineering

1Automate the data movement

Custodial and bank data moved to automated ingestion, eliminating the re-keying step that generated most downstream errors.

2Reconcile continuously, not at quarter-end

Reconciliation shifted to a rolling cadence with exception queues, so breaks surfaced within days of occurring rather than weeks later. Quarter-end became a confirmation, not an investigation.

3Templatise the reporting packages

Family packages were standardised to a common core with defined variations, so assembly stopped being bespoke craftsmanship and became a repeatable production step.

4Document the entity knowledge

The institutional knowledge trapped in one person's head was written down — entity by entity — removing the single point of failure and letting work be shared.

04The outcome

The close went from six weeks to nine days, and the character of the work changed more than the calendar did. The team moved from chasing data to interpreting it, families received timely numbers, and the office could take on additional relationships without adding the same effort linearly. The automation mattered, but the deeper win was turning a heroic, undocumented process into an ordinary, repeatable one.