Financial Close Automation for Banks With Multi-Entity Consolidation
Shortening a Close That Is Mostly Spent Waiting
Ask a finance team why the close takes nine days and the answers describe waiting rather than working. Waiting for a subledger to complete its cycle, for a reconciliation to be performed, for an intercompany balance to be agreed with another entity, for a journal to be approved, for a consolidation run that cannot start until every entity has submitted. Very little of the elapsed time is computation, and almost none of it is analysis.
That makes financial close automation banking groups need less a matter of faster processing and more a matter of removing dependencies. The reconciliation that blocks day four can usually be performed continuously. The intercompany break discovered on day six could have been caught the day it arose.
Why does the close take as long as it does?
Because it is a dependency chain of manual tasks, most of which are waiting rather than working.
| Close activity | Typical position | Could it move? |
|---|---|---|
| Subledger cycle completion | Day 1 to 2 | Partly, with earlier cut-offs |
| Account reconciliation | Day 2 to 4 | Yes, continuously before period end |
| Intercompany agreement | Day 3 to 6 | Yes, with continuous matching |
| Accrual and provision journals | Day 3 to 5 | Partly, with automated calculation |
| Manual correcting journals | Throughout | Should be eliminated at source |
| Currency translation | Day 4 to 6 | Yes, automated with recorded rates |
| Consolidation and elimination | Day 5 to 7 | Yes, once inputs arrive earlier |
| Review, certification, sign-off | Day 6 to 9 | Partly, with continuous certification |
| Statutory and regulatory reporting | After close | Yes, from consolidated data |
Reading the third column is the whole programme. Most activities can move earlier or out of the close entirely, and the ones that cannot are review and judgment, which is what a close should actually consist of.
How much of your close duration is computation, and how much is waiting for a dependency?
What is the single biggest lever?
Moving reconciliation out of the close.
A reconciliation performed continuously produces a break the day it arises, with an obvious cause and a routine fix. The same reconciliation performed at period end produces a break under time pressure with several weeks of candidate explanations, on the critical path, blocking everything behind it. Continuous reconciliation between subledgers and the ledger, between systems, and between entities converts the close from a discovery exercise into a confirmation exercise, which is where the days come from. The reconciliation-time argument is well made in this account of cutting close reconciliation time, and the subledger design that enables it in general ledger and subledger modernisation.
What does close orchestration add?
Dependency management, ownership, and a measurable critical path.
Most closes are run from a spreadsheet listing tasks with owners and target days. Orchestration replaces that with modelled tasks carrying predecessors, owners, status, evidence, and duration, which produces three things a spreadsheet cannot: a visible critical path, automatic notification when a predecessor completes, and historical data on where time is actually spent. That last item matters most, because close improvement without duration data per task is guesswork, and the tasks people believe are slow are frequently not the ones on the critical path.
What should be automated first?
Task status capture and dependency triggering, before any individual task is automated.
Automating a task that is not on the critical path shortens nothing. Instrument first, find the path, then automate along it. The typical finding is that a handful of reconciliations and one intercompany agreement determine the duration, and that automating a widely disliked but off-path task would have delivered no improvement at all.
Why is intercompany elimination so troublesome?
Because both sides must agree in amount, currency, and period, and mismatches arise easily.
Intercompany balances break for prosaic reasons: one entity books in a different period, the two use different exchange rates, a charge is recorded gross by one side and net by the other, or a transaction is recorded by one entity and not yet by the other. Each is small and each blocks elimination. Fix it with continuous matching rather than period-end agreement: match intercompany transactions as they are booked, alert both sides to a mismatch immediately, and require agreement before period end rather than during close. Then set the policy questions in advance, particularly which entity's rate governs and how timing differences are treated, since arguing about that at close is what turns a break into a two-day delay.
What does consolidation require beyond arithmetic?
Control assessment, uniform policies, translation, elimination, and minority interest.
IFRS 10 Consolidated Financial Statements, issued in May 2011, defines the principle of control and establishes control as the basis for consolidation, requiring a parent that controls one or more subsidiaries to present consolidated financial statements and setting out how to apply the control principle to identify whether an investor controls an investee. For a platform that means the entity structure and the basis for consolidating each entity is data to be maintained rather than a static configuration, and it changes with acquisitions, disposals, and changes in control. Model the group structure with effective dating, hold the consolidation basis per entity, and make the structure auditable, because a consolidation whose entity basis nobody can evidence is a finding waiting to happen. Where entities report under different frameworks, the alignment work is the subject of this guide to GAAP and statutory reporting differences.
Is your group structure and consolidation basis held as effective-dated data or as configuration nobody has reviewed?
How should currency translation be handled?
Automated, with rates recorded on the translation and the result reproducible.
Translation is straightforward arithmetic surrounded by decisions: which rate for which balance type, which date, which source, and how translation differences are presented. Automate the mechanics, record the rate applied and its source on every translation, and keep the calculation reproducible so a prior period can be re-derived. Manual rate entry during close is a recurring source of error and of unexplainable movements, and it is entirely avoidable with a rate feed and a documented policy per balance category.
How do you reduce manual journals?
By treating each one as a defect with a source, rather than as a workload to automate.
Manual journals at close are mostly corrections: an accrual the subledger did not compute, a misposting from a product system, a reclassification because the chart lacks a dimension, a rate applied incorrectly upstream. Automating the journal preserves the underlying defect and makes it permanent. Instead, categorise journals by root cause, fix the largest categories at source, and track journal volume as a quality metric rather than as a throughput one. The ones that remain should be genuine judgment entries, which are few and worth reviewing properly. Where the underlying figures differ between functions, the problem is upstream, which is the pattern described in reporting the same thing differently.
What is certification, and why does it matter?
Named sign-off with evidence attached, which turns a control into an auditable record.
Certification means a named person confirming that a reconciliation is complete and correct, or that a balance is supportable, with the evidence attached at the point of sign-off. It matters for two reasons. Auditors and regulators want to see the control operating rather than asserted, and internally it establishes accountability that a checklist tick does not. Implement it continuously rather than at close: a reconciliation certified when performed is a certification you already hold at period end, rather than another close-week task. The regulatory reporting that follows depends on this evidence chain, and the calendar discipline is the subject of this guide to regulatory filings calendars.
How do you handle the reporting that follows the close?
By generating it from consolidated data rather than assembling it separately.
Statutory accounts, regulatory returns, and management reporting all derive from the same consolidated figures, and assembling each separately creates the situation where three reports disagree and nobody can say which is right. Generate from one consolidated dataset with the reporting basis as a parameter, retain the submitted output, and reconcile any manual adjustment made in a report back to the ledger. That also removes a large block of post-close work, which is frequently as long as the close itself and rarely measured.
How should the programme be sequenced?
Instrument, then move reconciliation earlier, then intercompany, then consolidation, then reporting.
| Phase | Duration | Deliverable |
|---|---|---|
| Close instrumentation | 1 to 2 months | Tasks, owners, dependencies, durations, visible critical path |
| Continuous reconciliation | 3 to 5 months | Automated matching outside the close with break workflow |
| Continuous certification | 2 months | Sign-off with evidence at the point of performance |
| Intercompany continuous matching | 2 to 3 months | Real-time matching, immediate mismatch alerting, agreed policies |
| Automated translation | 1 to 2 months | Rate feeds, recorded rates, reproducible results |
| Journal root cause programme | Ongoing | Categorised journals, upstream fixes, declining volume |
| Consolidation platform | 3 to 5 months | Effective-dated structure, elimination, minority interest |
| Reporting generation | 2 to 3 months | Statutory, regulatory, and management reporting from one dataset |
Instrumentation first, because every subsequent investment should be aimed at the critical path and nobody knows what it is until tasks are timed. Expect the first measurement to be uncomfortable and useful in equal measure.
Which metrics matter?
Close duration and critical path composition, reconciliation completed pre-close, intercompany breaks at period end, manual journal volume by cause, and post-close reporting time.
Report close duration alongside the composition of the critical path, since duration alone does not tell you where to invest. Track the share of reconciliations completed and certified before period end, which is the leading indicator of a shorter close. Count intercompany breaks outstanding at period end, targeting zero through continuous matching. Report manual journal volume by root cause with a downward trend, because that is a quality measure rather than a workload one. Measure post-close reporting elapsed time separately, as it is frequently the larger half. And track restatements and post-close adjustments, since a fast close that produces corrections afterwards has moved work rather than removed it.
Close automation is unusual in that the biggest gains come from doing things earlier rather than faster. Reconcile continuously, agree intercompany balances as they arise, certify at the point of performance, and fix the upstream defects that generate manual journals. What remains is review and judgment, which is what the close was supposed to be about.
Frequently Asked Questions
Why does the close take as long as it does?
Because it is a dependency chain of manual tasks. Most of the elapsed time is waiting for reconciliations, journals, and approvals rather than computing anything.
What is the single biggest lever on close duration?
Moving reconciliation out of the close. A reconciliation performed continuously is a routine fix, while the same reconciliation at period end is a blocker on the critical path.
What does close orchestration actually add?
Visibility and dependency management. Tasks with owners, predecessors, and status turn a spreadsheet checklist into a critical path you can shorten deliberately.
Why is intercompany elimination so troublesome?
Because both sides must agree in amount, currency, and period. Mismatches arise from timing, FX rates, and inconsistent booking, and they surface at the worst moment.
What does consolidation require beyond adding entities together?
Control assessment, uniform accounting policies, currency translation, intercompany elimination, and minority interest treatment, which is why IFRS 10 defines control as the basis.
Why should journals be reduced rather than automated?
Because most manual journals correct something upstream. Automating the correction hides the defect, while fixing the source removes both the journal and the risk.
What is certification and why does it matter?
Named sign-off that a reconciliation or balance is correct, with evidence attached. It converts a control from an assertion into an auditable record with accountability.
How much can close duration realistically fall?
Substantially, and mostly by removing waiting rather than working faster. Continuous reconciliation, earlier subledger cut-offs, and fewer manual journals do most of the work.
