The close, compressed
Every finance team knows its close date. Fewer can say which task sets it. Month-end is a chain of dependencies — bank before receivables, receivables before intercompany, intercompany before consolidation — and the books close when the longest chain finishes, not when the busiest team does.
So automation only buys days where it touches that chain. Rules and integrations take out the mechanical hours; AI moves matching and accruals into the month, before the close window opens, and drafts the commentary for FP&A to edit. Switch it on task by task and watch where the date is decided — until the only thing left setting it is the review a person has to sign.
Three cases
Across Europe, the UK and the Americas. Intercompany, consolidation and eliminations grow with the group; a single entity has none.
Close lands
WD7
working day, by hand
Effort removed
0 h
of 148 person-hours a month
Critical path
9 tasks
end to end, with no slack
Constraint now
Intercompany matching
Group reporting · 9 h
Verdict
Run by hand, the close lands on working day 7.
The constraint now is intercompany matching, the longest step still done by hand on a nine-task critical path. Automating it next brings the close to working day 6.
Close calendar · 6 entities
Switch a task to automate it- Critical path
- Has slack
- Automated
- By hand
Task by task
Hours are elapsed · 8-hour working days from 09:00| Task | Lever | Hours | Starts | Ends | Slack |
|---|---|---|---|---|---|
| Bank reconciliationTreasury | AI Feeds matched daily; only exceptions reach a person | 8 h | WD1 09:00 | WD1 17:00Critical | Critical |
| Receivables & revenueReceivables | Integration Billing synced to the ledger; schedules post | 7 h | WD2 09:00 | WD2 16:00Critical | Critical |
| Payables cut-offPayables | AI Invoices captured, coded and matched on arrival | 9 h | WD1 09:00 | WD2 10:006 h slack | 6 h |
| AccrualsGeneral ledger | AI Proposed from orders, contracts and history before cut-off | 6 h | WD2 10:00 | WD2 16:0012 h slack | 12 h |
| Payroll journalsPayroll | Integration Payroll mapped straight into the ledger | 4 h | WD1 09:00 | WD1 13:0023 h slack | 23 h |
| Fixed assets & prepaymentsGeneral ledger | Rules Schedules post themselves in-month | 4 h | WD2 10:00 | WD2 14:0014 h slack | 14 h |
| Intercompany matchingGroup reporting | AI Matched continuously across entities and currencies | 9 h | WD2 16:00 | WD3 17:00Critical | Critical |
| FX revaluationGeneral ledger | Integration Rates pulled and revaluation run on schedule | 3 h | WD4 09:00 | WD4 12:00Critical | Critical |
| ConsolidationGroup reporting | Rules Consolidation engine replaces the workbook | 6 h | WD4 12:00 | WD5 10:00Critical | Critical |
| EliminationsGroup reporting | Rules Generated from matched intercompany pairs | 4 h | WD5 10:00 | WD5 14:00Critical | Critical |
| Tax provisionTax | Rules Provision pre-computed; the specialist reviews | 6 h | WD4 12:00 | WD5 10:007 h slack | 7 h |
| Flash P&LFP&A | Rules Built straight from the consolidated ledger | 3 h | WD5 14:00 | WD5 17:00Critical | Critical |
| Management packFP&A | AI Variance commentary drafted by AI, edited by FP&A | 8 h | WD6 09:00 | WD6 17:00Critical | Critical |
| Review & sign-offController & CFO | None: judgement stays with people | 6 h | WD7 09:00 | WD7 15:00Critical | Critical |
| Person-hours | 148 h | All by hand | |||
Notes on the model
What it shows
- Automation off the critical path saves effort, not days. The two headline numbers move independently, and most business cases only quote the first.
- The constraint migrates: intercompany matching first, then the management pack, and finally review and sign-off — the floor no tool removes.
- Group structure matters as much as tooling. Add entities and intercompany, consolidation and eliminations stretch faster than anything else.
How it works
- Fourteen tasks, each with an owner, an elapsed time, a crew and its dependencies. A forward pass gives earliest start and finish, a backward pass gives slack, and the constraint is the longest step on the critical path still done by hand.
- Each lever shortens its task. Where it lets the work run in-month — accruals, schedules, intercompany — it also removes the wait on cut-off.
- Effort is elapsed time multiplied by crew. The annual value prices the hours removed at an illustrative fully loaded hourly rate.
- Entity-level work runs in parallel, so elapsed time creeps up with group size while person-hours grow almost linearly. Group-level work grows with the web of intercompany flows.
What it leaves out
- Durations are illustrative, calibrated for a six-entity group on 8-hour working days. Your own close calendar is the only input that matters.
- Queueing, people juggling several tasks, late adjustments from the business and audit requests are all left out.
- Levers are on or off. In practice each one ramps, and each needs controls strong enough to trust its output before the review can shrink.
Next experiment · 02
Confidence, not magic
An AI categorises a live bank feed. You set the confidence bar — and own what slips under it.
Set the thresholdUsing this on a real decision?
Talk it through
The defaults are illustrative. The inputs that decide the answer are the ones you only know from inside the business.
Get in touch