Every close has the same shape. The first week is mechanical. The second week is investigation. Somewhere in there, someone finds a number that two systems disagree about, and the rest of the close bends around resolving it.
The instinct is to make close faster. Better checklists, earlier cutoffs, more people on it for five days.
That treats close as the problem. Close is not the problem. Close is when you find out.
Drift happens daily. Close is just the audit.
System drift is what happens when two systems that used to agree about the same transaction stop agreeing. A field gets renamed in the CRM. A billing cadence is set up against a different reading of the contract. An integration maps a value into a field that means something slightly different downstream.
None of that happens at close. It happens on an ordinary Tuesday, and it sits there.
By the time close surfaces it, the divergence has propagated into a forecast, an invoice, and a recognized revenue schedule. What was a field correction in week one is a reconciliation project in week five.
Why the usual fixes don't hold
Reconciliation reports. They compare two systems on a schedule, which means they find drift after it has already flowed downstream. They also require you to specify what to compare, so they only catch the disagreements you already thought of.
Tighter process. More approval steps slow the work down without watching the systems. Drift does not come from people skipping a step; it comes from two systems recording different parts of the same transaction correctly.
More headcount at close. This is the one being removed right now for efficiency, which is exactly why the underlying problem is about to get louder.
What continuous checking looks like
The alternative is not a faster close. It is a close with nothing left to find.
That takes a layer sitting above the systems of record that does four things without anyone opening it:
- Learns the structure of your data rather than requiring you to define it up front, and tracks how that structure changes over time. Bespoke contracts are structurally different from each other, so a schema-first approach breaks on the first real one.
- Establishes that records match. The opportunity, the obligation, the invoice schedule, and the recognized amount are the same transaction, resolved deterministically rather than fuzzy-matched.
- Detects divergence as it happens, statistically and always on, rather than on a monthly comparison run.
- Records lineage on every value. Actor, source, and timestamp, whether a person or an agent made the change.
A person still enters, at the end, on the exceptions, to make the judgment call. Not to do the checking.
What changes
Close stops being investigative. The exceptions were found and resolved in the week they occurred, when the fix was small and the person who made the change still remembered why.
And when the auditor asks how a number was produced, the answer is already assembled. Audit-ready is a state you hold, not a sprint you run in the fourth quarter.
