Who it's for: finance leads, controllers and accountants in SaaS or tech teams whose close takes eight to twelve working days and who want it shorter without losing control.
What you'll get: a way to find where the days go, eight changes that typically save the most time, a before-and-after worked example, where AI fits, and a one-slide pitch for your boss.
Reading time: about 12 minutes. Then pick one change for next month.
Why a shorter close matters (beyond bragging rights)
A 10-day close means leadership sees October's numbers in the middle of November. By then, the decisions those numbers should inform have often already been made. A long close also squeezes everything else: the team spends half of every month closing the last one, and analysis gets done in the gaps.
Going from 10 days to 5 is rarely about working faster. It's about waiting less, doing more before month-end, and agreeing what precision is actually needed by when.
Key terms
- Working day (Day 1, Day 2): working days after month-end. Day -1 is the last working day of the month.
- Critical path: the chain of dependent tasks that sets the earliest possible reporting date.
- Waiting time: time a task can't start because an input isn't ready. Usually the biggest source of lost days.
- Materiality threshold: the amount below which a difference or late item doesn't need adjusting this month.
- Auto-reversing journal: a journal set to reverse on the first day of the next month, so accruals don't need reversing by hand.
- Recurring journal: a journal posted automatically each month with the same or scheduled amounts, such as a prepayment release.
Step 1: find out where the days actually go
Most teams guess where the time goes and guess wrong. Before changing anything, time one close properly.
| Task | What "done" looks like | |
|---|---|---|
| ☐ | List every close task with its owner | One row per task in a spreadsheet |
| ☐ | Record start day, finish day and hours of actual work | Both elapsed time and effort captured |
| ☐ | Note what each late task was waiting for | A reason per delay: invoice, data, approval, person |
| ☐ | Mark the critical path | You can see which chain set the reporting date |
| ☐ | Total the waiting time vs the working time | A clear split between "doing" and "waiting" |
The common pattern: the close takes 10 days, but the actual work is about 4 days of effort. The rest is waiting for invoices, contract details, billing fixes, approvals or one busy reviewer.
AI shortcut: give an approved AI tool your task log (task, owner, start, finish, hours, waiting reason) and ask it to group delays by cause, show which delays were on the critical path, and rank causes by days lost. Check its grouping with the team; it only sees what's in the log.
Step 2: the eight changes that usually save the most time
Not every change suits every team. These are the ones that tend to move the reporting date, roughly in order of effort.
| Change | Typical saving | Effort | Where AI helps |
|---|---|---|---|
| 1. Move requests before month-end | 1–2 days | Low | Drafting reminder emails for each group |
| 2. Set and keep cut-off times | 0.5–1 day | Low | Converting close days into real dates |
| 3. Agree materiality thresholds | 0.5–1 day | Low | Listing items below threshold for a quick sign-off |
| 4. Automate recurring and reversing journals | 0.5 day | Low | Spotting journals that repeat each month |
| 5. Reconcile continuously, not at month-end | 1 day | Medium | Suggesting bank and payout matches daily |
| 6. Fix the billing-to-ledger sync | 0.5–1.5 days | Medium | Listing differences by customer and invoice |
| 7. Spread the review | 0.5–1 day | Medium | Flagging unusual journals so review focuses on risk |
| 8. Template the reporting pack | 0.5–1 day | Medium | Drafting variance commentary from the numbers |
Savings overlap, so they don't simply add up. The point is that five or six of these together often take a 10-day close to 5 or 6.
1. Move requests before month-end
Ask for accrual inputs, contract changes, starters and leavers on Day -3, not Day 2. Most of the information exists before the month ends. People just aren't asked for it.
2. Set and keep cut-off times
Give every external input a day and a time. Late items go to next month unless they're above the threshold. See the close calendar guide for how to set this up.
3. Agree materiality thresholds
Agree in writing, with your manager or finance lead, an amount below which a late invoice or a small reconciling difference isn't adjusted in the month. Many teams spend a day chasing items that couldn't change a single decision.
4. Automate recurring and reversing journals
Depreciation, prepayment releases, rent and standard recharges can be set up as recurring journals. Accruals can be set to reverse automatically on the first day of the next month. Both remove manual steps and a common source of errors.
5. Reconcile continuously
Reconcile bank and payment processor accounts weekly or daily during the month. At month-end, only the last few days are left to match.
AI shortcut: many accounting tools now suggest bank matches automatically. Spend ten minutes a day accepting the confident matches and clearing exceptions, rather than an afternoon on Day 1. A person still reviews every exception and signs off the reconciliation at month-end.
6. Fix the billing-to-ledger sync
If the billing system and the ledger disagree every month, the revenue close stalls while someone finds out why. Track the causes for two or three months, then fix the top one at source (a mapping, a missing tax code, a sync that fails on credit notes).
7. Spread the review
If one controller reviews everything on Day 4, Day 4 is the bottleneck. Review accounts as they're finished, from Day 2 onwards, and agree which low-risk accounts get a lighter review.
8. Template the reporting pack
Link the pack to the ledger or a reporting tool so numbers update when the ledger is locked. Then the last day is spent on commentary, not on copying numbers into slides.
AI shortcut: give AI the budget vs actual table and last month's commentary, and ask for a first draft in the same style for every variance over your threshold. Add the business reasons it can't know, and check every number it quotes against the pack.
Step 3: change one or two things a month
Changing everything at once breaks the close. A common approach is one or two changes per month over a quarter, measured each time.
| Task | What "done" looks like | |
|---|---|---|
| ☐ | Pick the change that removes the most critical-path waiting | One change chosen, with an owner |
| ☐ | Agree it with anyone outside finance it affects | Sales, HR or budget holders know what's changing |
| ☐ | Run the next close with the change | Change in place and tracked |
| ☐ | Compare planned vs actual days | You know whether it worked |
| ☐ | Keep it, adjust it or drop it | Decision recorded in the close notes |
Worked example: MonthEndClub Co Ltd goes from 10 days to 5
MonthEndClub Co Ltd is a fictional UK SaaS company with a four-person finance team. The team timed its September close. It took 10 working days, but only about 3.5 days of it was actual work on the critical path. The rest was waiting.
| Cause of delay on the critical path | Days lost | Change made | Days lost after |
|---|---|---|---|
| Waiting for supplier invoices and accrual inputs | 2.0 | Requests on Day -3; invoice cut-off at 15:00 on Day 1 | 0.5 |
| Billing system and ledger disagreeing | 1.5 | Fixed credit note sync; daily sync check | 0.5 |
| Bank and Stripe matching all on Day 1–2 | 1.0 | Weekly reconciliations during the month | 0.0 |
| Controller reviewing everything on one day | 1.0 | Review spread from Day 2; light review for low-risk accounts | 0.5 |
| Rebuilding the pack in slides | 1.0 | Pack linked to the locked ledger; AI-drafted commentary | 0.0 |
| Total waiting | 6.5 | 1.5 |
The close fell from about 10 days to about 5. Working time stayed at roughly the same 3.5 days, and waiting fell from 6.5 days to 1.5. 3.5 + 1.5 = 5 days.
One automated journal, set up once
MonthEndClub Co Ltd pays £24,000 upfront on 1 October for a 12-month software subscription. It used to calculate and post the release by hand on Day 3. Now it sets up a recurring journal for 12 months.
£24,000 ÷ 12 = £2,000 per month.
On payment (1 October):
| Account | Dr | Cr |
|---|---|---|
| Prepayments | £24,000 | |
| Bank | £24,000 |
(VAT on the invoice is left out here to keep the example simple.)
Recurring monthly journal (October to September):
| Account | Dr | Cr |
|---|---|---|
| Software costs | £2,000 | |
| Prepayments | £2,000 |
After 12 months, 12 × £2,000 = £24,000 has been released and the prepayment is nil. The month-end task changes from "calculate and post" to "check the recurring journal ran and agrees to the prepayment schedule". That's a small saving each month, but a dozen like it frees up most of a day.
Common mistakes and audit watch-outs
- Cutting reconciliations to save time. A faster close that skips controls isn't faster; it moves the work to the audit.
- Changing everything at once. If five things change in one month and the close goes wrong, you won't know which one caused it.
- Thresholds nobody wrote down. An agreed threshold is a control. An unwritten one looks like a missed adjustment.
- Recurring journals left running. Review recurring journals each quarter. A contract that ended in June still releasing in August is a common find.
- Measuring effort, not elapsed time. The reporting date is set by the critical path. Saving hours off the path doesn't move it.
- Using AI drafts without checking. AI can misread a number or invent a reason for a variance. Every AI output is checked against the source.
- Putting financial data into unapproved AI tools. Only use AI tools your company has approved for financial, customer or payroll data.
Tools that can help
Accounting software such as Xero, QuickBooks or Sage supports recurring and auto-reversing journals and suggested bank matches. Billing tools such as Stripe or Chargebee sync invoices to the ledger, and are worth reconciling daily. Reporting tools such as Fathom or Syft, or a well-built Excel or Google Sheets pack, can link to the ledger so numbers update automatically. Tools don't shorten the close on their own; fixing the waiting does.
Take it to your boss: one slide
Getting from a 10-day close to 5
The problem: our close takes about 10 working days, but only around 3.5 days of it is work. The rest is waiting for inputs, fixing data and queuing for review.
Three changes
- Ask earlier: accrual inputs and contract changes requested on Day -3, with fixed cut-off times.
- Automate the routine: recurring and auto-reversing journals, weekly bank and payout matching, fixed billing sync.
- Review as we go, with AI first drafts: review from Day 2; AI flags unusual journals and drafts commentary; people review and sign off.
| Risk | Control |
|---|---|
| Late items missed | Written materiality threshold, approved by finance lead |
| Recurring journals out of date | Quarterly review of all recurring journals |
| Weaker review | Risk-based review list; high-risk accounts reviewed in full |
| AI errors or data leaks | Approved tools only; every AI output reviewed |
How we'll measure it: working days to issue the pack, number of post-close adjustments, and reconciliations signed off by Day 4.
Key takeaways
- Time one close properly before changing anything. Most lost days are waiting, not working.
- Focus on the critical path. Saving time elsewhere doesn't move the reporting date.
- The biggest wins are usually earlier requests, cut-off times and written thresholds.
- Automate recurring and reversing journals, and reconcile during the month.
- Change one or two things a month, and keep the controls in place.
FAQs
Is a 5-day close realistic for a small SaaS team?
For many single-entity SaaS companies with standard billing, yes. Multi-entity groups, complex revenue or heavy manual billing usually take longer, and a 6 to 8 day close may be a sensible target.
Will a faster close mean more errors?
It can, if speed comes from skipping steps. If it comes from removing waiting, automating routine journals and agreeing thresholds, error rates often fall. Track post-close adjustments to check.
Where does AI save the most time in the close?
Typically in matching (bank, payouts, billing vs ledger), first-pass accrual lists, flagging unusual journals and drafting variance commentary. A person still reviews and signs off everything, using only company-approved AI tools.
What's a sensible materiality threshold?
It depends on the size of the business and who uses the numbers. Many teams set it as a fixed amount or a small percentage of monthly costs, agree it with the finance lead, and review it each year.
Related guides
- How to Build a Month-End Close Calendar (Day -3 to Day +5)
- What to Automate First in Your Close
- Reconciling Your Billing System to the Ledger
- Soft Close vs Hard Close, and What Changes at Quarter-End
More in Month-End Close guides.
From the Club
Five days back. Spend one of them outside. The Washed Cap · The Team Mug
Join the Club. New guides, early access and members-only batches. Join here.
This guide is general information about common practice. It isn't tailored to any business, and standards, tax rules and figures change. Check your own circumstances, the latest rules and, where needed, a qualified professional before acting.
Last reviewed: October 2026.