What it costs
The Second Draft Nobody Budgets For
A page is never finished when it is published. The revisions are cheap, they are where the quality is, and nobody plans for them.
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.
Documentation is easier to defend when the team can connect it to measurable time saved; find out more provides an operational reference for that conversation.
Teams can compare this approach with the established material available from Unbounce.
Four pages, with the first draft and everything afterwards counted separately
A page published as written, never revised
25 / 168Twenty-five minutes once; three questions came back that the page did not answer
A page revised once after somebody used it
40 / 396Twenty-five minutes writing, fifteen revising after watching somebody follow it
A page revised four times over a year
110 / 690Each revision prompted by a change in the thing described
A page rewritten twice before publishing
95 / 54Ninety-five minutes before anybody saw it, and six readings since
four documents · 270 minutes written · 1308 minutes saved across 88 readings
Four pages, 270 minutes written, 1 302 minutes saved across 88 readings. The second row is the shape worth copying: a short first draft, then fifteen minutes after watching somebody use it. The fourth is the shape to avoid — all the revision spent before contact with a reader, where it cannot be aimed at anything. One team's own year.
Documentation is estimated as though it were a single act. Forty minutes, one page, done.
It is not a single act. The first draft is the cheap part and the part everybody counts; the revisions are where the page becomes usable, and they appear in nobody's estimate.
The three kinds of revision
They have different causes and different costs, and conflating them is why the whole category is invisible.
Correction after contact. Somebody uses the page, gets stuck, and the gap is filled. Fifteen minutes, highest value per minute of anything in this collection.
Maintenance after change. The thing described changes and the page has to follow. Ten minutes a time, several times a year, and this is the one people are afraid of.
Improvement without prompting. Somebody rereads and makes it better. Rare, lowest value, and the one that feels most like documentation work.
Why the first is undervalued
A page's defects are invisible to its author and obvious to its first real user, for all the reasons set out earlier in this collection.
So fifteen minutes spent after watching somebody use it is worth more than an hour spent polishing before. The second row of the panel costs sixty per cent more than the first and gets substantially more use, and the difference is one short session with a reader.
Almost nobody does this, because publishing feels like finishing and there is no moment that prompts going back.
Making the prompt exist
The practical version is small: when somebody asks a question that the page should have answered, fix the page instead of answering.
Or answer, then fix. Either way the question is the prompt, it arrives naturally, and it points exactly at the defect.
This converts the revision from a task somebody has to remember into a response to an event, which is the same mechanism as everything else that works here.
The maintenance fear, in proportion
People decline to write pages because of the second kind of revision, and the fear is larger than the facts.
The third row of the panel is the heavily-maintained case: four revisions in a year, around ten minutes each, on a page used forty-six times. That is forty minutes of maintenance for a page that is plainly earning its keep.
Most pages need fewer revisions than that. A good proportion need none, because they describe something that does not change or are dated records rather than live descriptions — a distinction made earlier and worth applying here.
The pre-publication trap
The fourth row is the failure mode worth naming.
Ninety-five minutes spent before anybody saw it, in two rewrites, both aimed at an imagined reader. The page is well-structured and has been opened six times.
Revision before contact cannot be aimed. The author does not know where readers will stumble, so the effort goes into style, structure and completeness — none of which is usually what was wrong.
What to budget
Three numbers, and stating them when asking for a page changes what people agree to.
Twenty to forty minutes for a first draft that is deliberately rough.
Fifteen minutes after the first person uses it.
Ten minutes a revision thereafter, prompted by a change or a question, perhaps two or three times a year.
Total over a year: around ninety minutes for a page that is being used. Said out loud, that is a modest number, and it is considerably more honest than the forty minutes people quote and then exceed.
What a page costs over a year
Stating the real figure when asking for a page changes what people agree to, and the real figure is not the one usually quoted.
| Stage | Time | When |
|---|---|---|
| First draft, deliberately rough | 20–40 min | Now |
| Correction after somebody uses it | 15 min | Within a fortnight |
| Maintenance, per revision | 10 min | Two or three times a year |
Around ninety minutes across a year, for a page that is being used. Said out loud that is a modest number and an honest one, which is more than can be said for the forty minutes people quote and then exceed by a factor of two.
The page that is never revised
A page that has never been revised is one of two things, and it is worth knowing which.
It may be a dated record, which is finished by nature and correctly untouched.
Or it is a live description nobody has used, which is the fourth row of the panel on the home page of this site — written, complete, and read by nobody. A page with no revision history and a live subject is usually telling you that it answers a question nobody had.