A renewal posts to the budget and nobody in IT recognizes the line. The tool it pays for has been running since someone needed it for a week, and that week ended a long time ago.
The work behind that line was real. It took real hours, it created an obligation that renews on its own, and it never appeared in anything leadership was tracking. By the time an invoice makes it visible, the decision that created it is months old and the capacity it consumed is already spent.
I get called in after a few of these land at once. What I'm usually asked for is better reporting. That isn't the problem. The problem is that work can enter the system without being recorded, owned, or placed in line with everything else, so the plan describes a smaller organization than the one that's actually doing the work.
There is more than one way in
There's usually a front door. It has an intake record, a named owner, some agreement about what finished means, and at least a chance of being forecast.
Then there are the other ways in. A message that starts with "can you jump on this." A request routed straight to the engineer who already knows the system. A vendor added for a short trial that nobody's set an end date on. A senior leader who says get it moving, which the team reasonably hears as permission to skip the definition step.
These aren't rare exceptions and they aren't evidence of bad intent. They persist because the official route is slower than the need, or incomplete, or shaped for a kind of request that doesn't match what's being asked. People use the path that works. If the official path worked better, they'd use it. The behavior is a reading of the system, not a character trait.
What matters is the arithmetic. If it consumes capacity, it counts, whether or not anything records it.
Invisible work doesn't change the plan, and that's the whole problem
A commitment made in the plan assumes a certain amount of the team's time is available to meet it. Work that arrives outside the plan takes some of that time. The plan doesn't know, so it doesn't move. Nothing gets renegotiated, no date changes, and the shortfall accumulates quietly until several commitments miss in the same week.
That's why the failure looks sudden when it isn't. Leadership experiences it as everything slipping at once with no obvious trigger. What actually happened is a long sequence of small unrecorded consumptions, each one defensible on its own.
Spend behaves the same way. A trial that nobody's watching keeps renewing. Usage that was described as temporary becomes the baseline for a true-up. A one-time fix from a vendor turns into steady-state cost because no one owned the exit. None of this requires anyone to be careless. It's what happens when work can ship without a record that has an owner and an end.
A ledger is a work record, not a finance artifact
The word ledger makes people think of accounting. It isn't that. It's a plain list of the work that entered, with enough on each line to own it and revisit it later.
Six fields carry most of the weight:
- who asked and who benefits, which frequently aren't the same person;
- who owns it through to the end, including whatever cost continues after it closes;
- what finished means, stated concretely enough to check;
- rough effort or cost, at the order of magnitude a person can give in a minute;
- a review or expiry point, where the work creates something that keeps running;
- what got delayed, paused, or reduced because this entered.
The fields aren't universal. Which ones you need depends on how the work behaves in your organization, how much of it there is, and how quickly decisions get made. A short list that people actually complete beats a longer one they route around.
That last field does more work than the rest of them combined. If an exception enters and nothing is recorded as moving, the displacement still happened. It just landed on whichever commitment was least able to defend itself.
Recording it is only half the route
A record makes the work visible. It doesn't yet make it comparable. Work that has a ledger line but sits outside the ordering everything else competes in is still, functionally, a side door with better paperwork.
Routing is complete when the item enters the same sequence as the rest of the work and can be reordered against it. That's a separate mechanism with its own failure modes, and I've written about it in The One Queue Rule.
If work consumes capacity, it has to enter the same visible ordering as everything else. Otherwise the delay has already been created, and the only thing you've achieved is that nobody can see it yet.