Asking
The Answer That Already Exists Somewhere
A large share of the questions crossing a gap have been answered before. The cost is not the asking; it is that the answer gets rebuilt differently.
three things · six nights
Three questions, six nights, and two of them avoidable.
Six questions from one week where the answer had already been given once
Which of the two approaches did we settle on?
Settled six weeks earlier in a thread nobody could find; re-argued from scratch
What is the retention period for the export files?
Written in the runbook; the asker did not know the runbook covered it
Did we agree to support the old format?
Agreed in a call, never written down, and two people remembered it differently
Who signs off on price changes?
Written down, but in a document last updated before a reorganisation
What was the reason we moved off the nightly job?
The decision was recorded and the reasoning was not
Is the staging data refreshed automatically?
six questions · nine round trips beyond the first
Six questions, eleven round trips beyond the first, and five of the six had been answered before. Two were recorded and unfindable, two were recorded without the reasoning, one was never recorded at all. Nothing here divides effort by outcome; the point is where the answers went. One team's own week, traced afterwards.
Teams with a wide gap ask the same questions repeatedly, and the repetition is rarely noticed because the two askings are weeks apart and usually by different people.
The best response to The Answer That Already Exists Somewhere is usually better evidence, not tighter supervision. project time tracking gives teams a transparent way to review hours and activity while leaving interpretation to the people responsible for the work.
For broader context, ClickUp's project platform offers an official example of the tools teams use around this kind of process.
The cost is not the night. It is that the answer gets reconstructed rather than retrieved, and reconstructions differ.
Why it is worse than wasted time
A question answered from memory produces an answer that is probably right. Two questions answered from memory six weeks apart produce two answers that are probably right and may not match.
Now there are two versions of what was decided, both held with confidence, and the disagreement surfaces later in the form of work built on the wrong one. That failure is expensive in a way the original night was not, and across a gap it takes several crossings to untangle, because each person is defending a memory rather than pointing at a record.
This is the real argument for writing decisions down, and it is a stronger argument than the usual one about institutional memory.
The four ways an answer goes missing
Never recorded. The decision was made in a call or in a conversation between two people and exists only in their heads.
Recorded without the reasoning. Somebody wrote down what was decided and not why. Six weeks later the decision looks arbitrary, somebody proposes the alternative, and the whole argument runs again — because the reasoning, not the conclusion, is what prevents reopening.
Recorded somewhere unfindable. It exists, in a thread or a document nobody thinks to search. Functionally identical to not existing, with the additional cost that somebody spends twenty minutes looking.
Recorded and stale. It exists, it is findable, and it has been superseded by something that was not recorded. This is the worst of the four because it produces confident wrong answers.
| How it is missing | What it costs |
|---|---|
| Never recorded | The answer is rebuilt, and two rebuilds differ |
| Recorded without the reasoning | The decision looks arbitrary and gets reopened |
| Recorded somewhere unfindable | Twenty minutes of searching, then a night |
| Recorded and stale | A confident wrong answer, which is the worst of the four |
What to write when a question is answered twice
The practical rule that catches most of this: if you find yourself answering the same question a second time, write the answer down somewhere permanent and say where you put it.
Not the first time. The first time it is a question. The second time it is a pattern, and the marginal cost of recording it is about two minutes against an unknown number of future nights.
This works because it requires no foresight. Nobody has to predict which answers will be needed again; the second asking is the prediction.
Writing the reason, not just the decision
A recorded decision without its reasoning is about half as useful and is the most common form of record.
"We use the nightly job" prevents nobody from proposing the alternative. "We use the nightly job because the streaming version could not guarantee ordering and ordering matters for the ledger" settles it, and it also tells a reader what would have to change for the decision to be worth revisiting — which is the thing a bare conclusion can never do.
Two clauses. The decision and the because.
Where it goes
Next to the thing, not in a general store of decisions.
A decision about the export job belongs with the export job. A decision about who approves price changes belongs wherever somebody looks when they want to change a price. The general repository of decisions is a structure that everybody agrees is a good idea and nobody searches.
The test is the same as everywhere else in this collection: will somebody with the question in front of them encounter this without having to think of looking for it?
Asking better when the answer might exist
From the asker's side, one sentence added to a question removes a whole category of second crossing.
"I have looked in the runbook and the channel history and could not find it" tells the recipient that pointing you at the runbook is not an answer, and stops the exchange where one person says "it is written down" and the other says "where".
It also, incidentally, means you look first, which resolves a decent fraction of the questions before they are sent at all.
Searching before asking, and its limits
The obvious instruction is to look before asking, and it is worth stating what it is reasonable to expect of that.
Five minutes. Somebody with a question should spend about five minutes looking in the two or three places where an answer of that kind would live, and then ask. Expecting more produces people who spend half an hour failing to find something and then ask anyway, having lost the half hour.
Where five minutes routinely fails, the problem is not the asker. It is that the records are in places nobody would think to look, and the fix belongs with whoever puts them there rather than with whoever cannot find them.