The month-end close as a design problem
Most finance teams treat the close as a reporting exercise. The fastest treat it as a design problem, and the first half of the improvement needs no new system at all.
- Slow closes are rarely caused by slow accountants. They come from dependencies nobody has defined, work that waits for month end without needing to, and rework caused by poor data upstream.
- The single most valuable fix is a close calendar that maps dependencies, with named owners and deadlines by day. It is a management tool, not a checklist.
- A large share of typical close work can be moved to before month end or removed altogether. A tiered approach to reconciliations and a written materiality policy for accruals do most of that work.
- Automate last. Automating a badly sequenced close gives you a badly sequenced close that runs faster in places that were never the bottleneck.
Ask a financial controller why the close takes twelve days and you usually get a list of people and systems. The plant teams send inventory data late. Bank statements arrive as PDFs. One senior accountant is the only person who can run the consolidation. Everything on the list is true, and none of it is the real reason.
The reason is structural. In most companies the close was never designed; it built up over time. Tasks exist because they existed last month. The order of tasks reflects who sat where in an office years ago. Nobody can say, for any particular task, what it holds up and what holds it up. A process with no explicit structure falls back on its slowest possible form, where everything waits for everything else.
The three places the time goes
When we map a close hour by hour, which we do before recommending anything, the elapsed time consistently breaks into three categories:
- Waiting. Tasks sit idle because a predecessor is not finished, or because nobody knew it was. In closes nobody manages, waiting regularly takes up more calendar time than the work itself. The consolidation analyst who starts on day eight could have started the group’s static entities on day two.
- Work that does not belong in the close. Vendor reconciliations, fixed asset registers, intercompany matching, and standard allocations end up in close week out of habit. Anything that does not need the month to be complete can be done before the month ends.
- Rework. This one does the most damage. Errors made upstream, such as miscoded purchases, dispatches never billed, and missing goods receipts, surface during the close and are fixed by the accounting team every month, without anyone recording where they came from. The close absorbs the cost of every process failure elsewhere in the company, and because it absorbs the cost, nothing upstream improves.
In a close that was never designed, more than half the elapsed time goes on waiting and rework
Illustrative breakdown of close time in a mid-market group, working days relative to month end
Illustrative figures. In the redesigned close, work that does not depend on the month being complete is done in the days before month end.
Source: SA&A analysis
Soni Aditya & Associates
The close absorbs the cost of every process failure in the company, and because it absorbs the cost, nothing upstream ever improves.
The close calendar, and why it carries the most weight
What separates fast closes from slow ones is not glamorous. It is a close calendar that works as a map of dependencies. That means more than a list of tasks. Every task carries four attributes:
- A named owner: a person, not a department.
- A deadline by day and time, such as “day 2, 2 p.m.”, counted from the last working day of the month so that the calendar survives weekends and holidays.
- Explicit predecessors: what has to finish before this task can start.
- A defined output: what “done” produces, and where it goes.
Building the calendar forces two conversations most finance teams have never had. Why does this task exist, and why does it wait for that one? In our experience, simply resequencing existing tasks, with no new systems and no new people, recovers a large part of the elapsed time. The later gains take harder work. The first ones come from structure alone.
Resequencing moves much of the close to before month end, with no extra people or systems
Timing of core close activities in a typical mid-market group, before and after dependency mapping
Illustrative. Durations and dependencies vary from group to group; the pattern reflects close redesigns in mid-market companies.
Source: SA&A analysis
Soni Aditya & Associates
Move work out of the close
Two policies move most of the effort out of close week.
Tiered reconciliations
Not every account needs the same rigour every month. A tiered policy gives high-risk accounts, such as bank, revenue interfaces, and intercompany balances, a full reconciliation monthly. Medium-risk accounts get a full reconciliation quarterly with a review of movements each month, and low-risk accounts get an analytical review. That is a better control and a faster one, because attention goes where the risk of misstatement actually is. The tiering itself should be documented and approved; auditors respond well to a deliberate policy and badly to inconsistency nobody can explain.
A materiality policy for accruals
Close teams lose whole days chasing precision on accruals that do not matter, such as hunting for the exact courier invoice behind a ₹40,000 estimate in a company with ₹80 crore of monthly expenses. A written policy sets a threshold below which standard accruals roll forward, an estimation method for each category of expense, and a timetable for truing up. The hunt becomes a fifteen-minute exercise with a documented basis. The goal of a close is reliable materiality, and absolute precision is not part of it.
Fix upstream data at the source, with a feedback loop
Rework only disappears when the close stops absorbing it without comment. The mechanism is a close defect log. Every error the accounting team corrects during the close is logged with its origin, meaning the process, the team, and the system. When the heads of those functions review the log each month, anecdote (“procurement coding is always wrong”) becomes data (“214 miscoded lines this month, 80 percent of them in two categories”). Within a quarter the pattern is visible, and within two the volume falls, because problems with a name and a count get fixed.
Automate last
Automation is the first instinct and should be the final step. Automating a badly sequenced close just runs the wrong process faster, and locks it in. The order that lasts is to map, resequence, move work out, and fix upstream data, and only then to automate what remains. By that stage the candidates are obvious and usually small: importing bank statements, recurring journal templates, intercompany matching, and assembling reports. Most of them are configuration work in systems the company already owns rather than new platforms.
What a good close looks like
The end state can be measured. A well-designed close in a mid-market group produces a flash P&L by day three and complete financials by day five or six. It also passes the harder test, which is to need very few post-close adjustments, because quality was built into the process rather than inspected in at the end. Three metrics are worth tracking: days to complete, the number of post-close adjustments, and recurring defects by source. When all three fall together, the close is really improving. When only the first one falls, corners are being cut, and the audit will find them.
A slow close is a design debt, and the whole company pays interest on it twelve times a year. Repaying it takes less time than most CFOs expect, and it starts with a calendar rather than a purchase order.
