In short. A UK-listed financial services group compressed its month-end timetable from ten working days to five over two close cycles. No general-ledger replacement, no consolidation-tool migration. The wins came from a sub-certification cascade aligned to UK CGC Provision 29 and a disciplined reconciliation library standardised across entities. The cost was roughly forty days of senior practitioner time spread across two months.

1. The setup

Mid-cap UK-listed group, financial services parent with around fifteen reporting entities, mixed ledger landscape: one major ERP at parent, a second platform at the largest subsidiary, three smaller boutique systems at acquired entities. Consolidation handled in a mature consolidation tool with reasonable but not excellent intercompany matching. Monthly close took ten working days, measured from period-end to board pack.

The board had asked for five days. The CFO’s instinct was to launch an ERP rationalisation programme. The CFO was right that the platform divergence was an annoyance and wrong that it was the bottleneck.

2. The diagnosis

The first three days of the engagement were spent watching the close happen. This is unglamorous but it is the only way to learn where time actually goes. The findings, in rough order of materiality:

  • Reconciliations were the bottleneck for days six through nine. Not the volume of reconciliations; the inconsistency of them. Every subsidiary used a slightly different format, a slightly different definition of “in-scope”, and a different threshold for what required senior review.
  • Sub-certification was sequential. The finance director at each subsidiary signed off only after every reconciliation had cleared, then the group financial controller signed off only after all subsidiary FDs had signed off. Each step waited for the previous one to finish completely.
  • Intercompany matching took an average of one and a half days. Not because the system was slow; because matching definitions diverged between entities. One entity recognised reinsurance premium on a cash basis, its counterparty recognised on an accrual basis, and the difference was reconciled manually each month.
  • The board pack was a bottleneck of its own. Final commentary was drafted only after all numbers had cleared, then reviewed by three layers before issue.

The ERP divergence was real but not the bottleneck. The bottleneck was that nothing happened in parallel because nothing had been designed to.

3. The first lever: a sub-certification cascade

The first structural change borrowed directly from the UK Corporate Governance Code Provision 29 requirement that the board confirms the effectiveness of risk management and internal controls. Sub-certification cascades are the practical implementation of that confirmation in financial reporting: each tier signs off only on what it owns, and the higher tier inherits that sign-off rather than reproducing it.

The redesign in practice:

  • Tier one: process owners at each entity certify reconciliations within their domain by close-day three. One certificate per process; one signature per certificate.
  • Tier two: subsidiary financial controllers certify completeness of tier-one certificates by close-day four. They do not re-perform reconciliations; they confirm that every required certificate has been received and signed.
  • Tier three: group financial controller and group CFO sign the consolidated certificate by close-day five, inheriting the tier-two confirmations and adding only the group-level reconciliations they directly own.

The legal mechanics matter here. The cascade does not transfer accountability upward; it documents accountability downward. The group CFO still signs the group certificate and still bears the responsibility. The cascade gives the CFO a written, dated, signed evidence trail that supports the signature.

4. The second lever: a reconciliation library

The second change was a standardised reconciliation library, owned at group, used by all entities. One template per reconciliation type. Standard format. Standard threshold definitions. Standard tier-one certifier role. Standard escalation path for reconciling items above threshold.

The library covered about thirty distinct reconciliation types: cash, intercompany, suspense, payroll, fixed assets, deferred tax, expected credit loss provisions, premium receivable, claims provisions, technical reserves, intercompany loans, fair-value level-3 movement, and so on. Each had a one-page template with named fields, prescribed sources, and a required certifier role rather than a named person.

The library was not a piece of software. It was a folder of templates and a written discipline. Three rules:

  • If a reconciliation does not appear in the library, it does not count for sub-certification.
  • If a reconciliation appears in the library but uses a non-standard format, it is rejected and resubmitted.
  • Threshold and certifier-role changes require group approval and a written change log.

Three months in, the library was the single most powerful artefact in the close. Auditors found their work easier; subsidiary teams found their work clearer; group found the file complete.

5. What we did not do

Three changes we considered and rejected, all of which the CFO had wanted at the start.

We did not migrate any general ledger. Platform divergence was an annoyance, not a blocker. A ledger migration would have cost the equivalent of three close cycles and produced no compression beyond what the cascade and library already delivered.

We did not implement any close-management software. Several vendors quoted six-figure annual subscriptions. The library and the cascade were a folder, a written instruction, and disciplined enforcement. After the timetable compressed, the question of close-management software became easier to answer because the underlying process was already standardised. A tool layered onto chaos amplifies chaos.

We did not change the consolidation tool. The existing tool was adequate; its limitations were not on the critical path.

6. The numbers

Cycle one, two months in: ten days to seven days. The cascade had partially landed; the library was about half-populated.

Cycle two, four months in: seven days to five days. Library complete; cascade fully enforced; intercompany matching definition standardised across entities.

The board pack moved earlier by an analogous discipline: pre-drafted commentary using rolling-forecast numbers as placeholders, replaced with actuals during the final two days and reviewed in parallel rather than in series.

7. The lessons that travel

Three observations that apply beyond this group.

First, technology spend is rarely the answer in a close-compression project because it is rarely the bottleneck. The bottleneck is usually documentation discipline and process serialisation. A well-designed cascade and a well-disciplined reconciliation library are cheaper and faster than any platform decision.

Second, the cascade is legally defensible only if the certificate trail is contemporaneous and signed. After-the-fact reconstruction is not sub-certification; it is paperwork. The discipline that makes the cascade work is the same discipline that makes the audit-evidence chain hold.

Third, compression past five days is a different problem. It usually requires changes to general-ledger automation, intercompany matching automation, and consolidation-tool capability. Those changes have real cost and real risk. Most groups do not need to go past five. The board pack benefit is concentrated in the move from ten to five; the next move, from five to three, costs more per day saved than the first one and delivers less governance benefit.

— DK Buledi, December 2025