What it costs
What the Forty Minutes Came Out Of
Writing does not take time from a reserve; it takes it from something specific that was already planned, and naming that thing is most of the cost conversation.
What nobody wrote down here, and what it cost
How the staging database is rebuilt
Asked of the same person seven times
Four minutes each, plus the interruption
Which client agreed to the export
Reconstructed from mailbox searches
Twice by people not on the thread
two things nobody wrote down · 10 occasions it cost somebody something
Two things, ten occasions, and nobody billed any of it to anybody.
Where the real question is capacity rather than willingness, this solution can supply the work-pattern evidence that a documentation discussion otherwise lacks.
The broader workflow can also be checked against guidance from Hotjar.
Four pages from one quarter, with what the writing displaced recorded alongside
The deploy runbook
95 / 372Two evenings; the author's own ticket slipped into the following week
A note written during a meeting
6 / 100Displaced nothing: it was typed while the author was already explaining
The onboarding page
240 / 360Four days, agreed in advance, and the team's sprint commitment was reduced to match
A design document nobody asked for
180 / —Three hours taken from a Friday afternoon, displacing nothing anybody missed
four documents · 521 minutes written · 832 minutes saved across 40 readings · one with no saving claimed
Four pages, 521 minutes written, 882 minutes saved across 40 readings. The second row is the one to look at: its cost was near zero because it was attached to something already happening. The third is the honest alternative, where the cost was real and was budgeted openly instead of absorbed. One team's own quarter.
"It only takes forty minutes" is the sentence used to argue that writing is cheap. It is accurate about the duration and misleading about the cost.
Nobody has forty unallocated minutes. The forty minutes come out of something, and the something is the actual price.
Naming the displaced thing
The useful question is not how long a page takes but what does not happen because of it.
For most people the answer is one of four things: a piece of their own work slips, a review waits, an interruption goes unanswered, or they finish later than they meant to.
The fourth is the most common and the worst, because it is the one option that requires nobody's agreement and appears in no record. Writing absorbed into somebody's evening is writing that the team received for free and that the author paid for alone.
Why this matters for how requests are made
A request to write something is a request to displace something, and most requests do not say so.
"Could you write that up?" asks for forty minutes without naming where they come from, which leaves the person asked to find them privately. The honest version names the trade: "could you write that up instead of picking up the next ticket?" — which is a question somebody can actually answer, and may answer with no.
Teams that make the trade explicit discover two things: that documentation gets written, and that the number of things worth writing is smaller than anybody said when it was free.
The three honest ways to pay
| How | Cost falls on | Visible to |
|---|---|---|
| Attached to work already happening | Nobody, almost | Nobody, which is fine |
| Budgeted openly, in place of other work | The team's output | Everybody |
| Absorbed into somebody's own time | One person | Nobody |
The first is the best and the rarest to engineer. The second is the honest fallback. The third is what happens by default and is the one that quietly ends careers as documentation writers.
Attaching it to something already happening
The second row in the panel is the ideal case and it is not an accident of personality.
Somebody was explaining a thing. Instead of explaining it in chat, they typed it into a page and sent the link. The explanation was going to happen either way; the marginal cost of making it durable was around ninety seconds.
Most of what a team could usefully write is like this. It is attached to an incident that is being handled, a question that is being answered, a decision that is being made — all moments where the material is already in somebody's head and the only extra cost is putting it somewhere that persists.
The fourth section of this collection is mostly about finding those moments.
Budgeting it openly
Where the writing genuinely requires a block — the onboarding page in the panel took four days — the honest route is to reduce what else is promised.
This is unpopular because it makes the cost visible at exactly the moment when somebody has to defend it. That visibility is the point. A four-day page that nobody will defend out loud is a four-day page that should not be written, and finding that out before the four days is a saving.
The evening absorption problem
The default route is the one nobody chooses and everybody takes.
It looks generous and it has three bad properties. The cost is unknown to the team, so the team cannot judge whether it was worth it. It falls on whoever is most willing, which is a poor allocation rule. And it is unsustainable in the specific sense that the person doing it stops after a year or two, usually without saying why.
A team whose documentation depends on this is not getting it cheaply. It is getting it on credit from one person.
Asking in a way somebody can answer
A request that does not name the trade leaves the person asked to find the time privately, which usually means their evening. Naming it makes the request answerable.
- "Can you write that up instead of picking up the next ticket?"
- "Can you spend an hour on this and we'll drop the smaller change from the sprint?"
- "Can you write four lines while you're explaining it, rather than a page later?"
The first two can be refused, which is the point: a cost that can be declined is a cost somebody has agreed to. The third costs almost nothing and is the one to reach for first.
Teams that start naming the trade find the list of things worth writing gets shorter, which is also a result.
What to ask before writing anything long
Two questions, thirty seconds.
What is this displacing? If the answer is "my evening", say so to somebody before starting.
Could it be attached to something already happening instead? If somebody is going to explain this anyway in the next fortnight, the page costs ninety seconds then and forty minutes now.
Those two questions do not reduce the amount a team writes. They change which hours it comes from, which is most of what determines whether anybody is still writing in a year.