Most elapsed time is waiting
62% of the close calendar is a step waiting on an input rather than a person working. Adding effort to a waiting step does nothing, which is why so much close optimisation fails.
Research · updated August 2026
Finance teams describe their close as slow in general terms. Timed properly, it is almost never general — it is two or three specific steps that everything else waits on, and the rest of the effort is spent optimising things that were never the constraint.
Tell us your entity count and close length. We will tell you where you sit and where the days likely are.
A representative close, timed. Waiting time is shown separately from working time because they respond to entirely different interventions.
What we found
These come from measuring each step rather than asking teams how long things take, which produces meaningfully different answers.
62% of the close calendar is a step waiting on an input rather than a person working. Adding effort to a waiting step does nothing, which is why so much close optimisation fails.
It gates the subledger tie-outs, which gate the trial balance, which gates everything else. Where it takes three days, the whole close inherits three days.
Two entities add about 30% to a single-entity close. Nine entities add roughly 180%, because intercompany creates dependencies between entities rather than just more work.
The last three days are usually spent waiting on someone outside finance — an accrual estimate, a headcount confirmation, a legal opinion on a contract.
Teams consolidating in Excel spend two to three days there regardless of entity count below about ten. It is one of the few steps that automation removes rather than shortens.
Teams closing in five days do not work more hours. They have fewer dependencies, cleaner reconciliation upstream, and no manual consolidation step.
The most useful distinction in close analysis is between a step that takes four hours of somebody’s effort and a step that sits for two days waiting on an input. They look identical on a calendar and respond to entirely opposite interventions.
Working time responds to automation and to headcount. Waiting time responds only to removing the dependency, and adding people to it makes the close more expensive without making it shorter. Teams that have tried to accelerate a close by hiring have usually been solving the wrong one.
In thirty-one of thirty-eight teams, bank reconciliation sat on the critical path. Not because it is difficult, but because it is upstream of the subledger tie-outs and therefore upstream of everything.
Continuous reconciliation removes it entirely rather than shortening it, which is why it is the single highest-leverage change available to most teams. A close that starts with the bank already tied begins two or three days ahead of one that does not.
Adding an entity does not add a fixed amount of close work. It adds intercompany relationships with every other entity, and each of those is a dependency that has to be matched and eliminated before consolidation can run.
That is why the curve bends upward. Groups at nine or more entities that still consolidate manually are usually the ones with the longest closes in the sample, and the consolidation step alone is frequently a third of their calendar.
The final days of a long close are rarely finance’s own work. They are waiting for somebody in operations to confirm an accrual, somebody in sales to confirm whether a deal closed, or somebody in legal to opine on a contract modification.
These are process problems rather than system problems, and they are the part a software purchase does not fix. The teams that solved them did so by moving the request earlier rather than by chasing harder — asking for the accrual estimate on day 25 rather than day 3.
Continuous bank reconciliation, subledgers that tie without a monthly exercise, no manual consolidation, accruals derived from data rather than estimated, and external inputs requested before period end.
None of those require heroics and all of them require the reconciliation to have already happened. The teams closing fastest are not the ones working hardest during the close; they are the ones for whom most of the close already occurred during the month.
Thirty-eight finance teams whose closes we timed at step level between 2024 and 2026, either during a health check or in the first month of an engagement. Timing is from system timestamps and observation rather than from teams reporting their own durations.
Working time and waiting time are separated by whether anyone was actively performing the step. A step where the owner was working on something else is counted as waiting even where the calendar says it was in progress.
The sample skews toward companies that had a close problem, since that is frequently why we were engaged. It is likely slower than the true mid-market median, and we would treat the eleven-day figure as an upper-middle estimate rather than a central one.
Entity counts range from one to fourteen, revenue from roughly $8M to $180M. No team is identifiable in the aggregates.
Questions
Two or three steps explain most closes. Optimising the others is effort spent off the critical path.