Counting
Putting Writing Into What People Are Judged On
If writing appears nowhere in how work is judged, it competes with work that does, and loses every time for perfectly rational reasons.
What nobody wrote down here, and what it cost
That the runbook saved somebody a morning
Nobody told the author
Six occasions, none reported back
That a note ended an argument
The author never heard
Both times from people who joined later
two things nobody wrote down · 8 occasions it cost somebody something
Two things, eight occasions, and the author heard about none of them.
Teams testing a new knowledge workflow can employee activity visibility to establish a baseline, then compare the result after ownership and review rules change.
A useful independent point of reference for this part of the process is Google Search Central.
One team before and after writing was named in the review criteria
The quarter before, by the one person who wrote
210 / 532Four pages, written voluntarily, mentioned in no review
The quarter before, by everybody else
0 / —Four people, no pages
The quarter after, by the same prolific person
180 / 465Three pages, and the writing appeared in their review for the first time
The quarter after, by everybody else
165 / 377Five pages, written by three of the four people who had written none
four documents · 555 minutes written · 1374 minutes saved across 98 readings · one with no saving claimed
Four rows, 555 minutes written, 1 329 minutes saved across 98 readings. The change in the last row is the finding, and the mechanism was not enthusiasm: three people who had never written anything were told, in advance, that it would be looked at. One team's own two quarters.
Everything in this collection so far has been about cost. This page is about the other half: whether the writing counts for anything once it is done.
In most teams it does not appear anywhere in how people are assessed, which means it competes against work that does and loses, correctly.
What the competition looks like
Somebody has a Thursday afternoon. They can ship a change that will be visible in the standup, reviewed by a colleague and present in their review, or they can write a page that appears nowhere.
They are not choosing between virtue and laziness. They are choosing between two uses of an afternoon, one of which is recognised by the organisation and one of which is not.
Asking people to choose the second repeatedly, with no change to the first, is asking them to accept a cost on behalf of colleagues they cannot name.
What naming it changes
The panel above is one team's two quarters. In the second, three people who had never written anything wrote something.
The mechanism was not motivation. It was that somebody said, in advance and specifically, that writing would be looked at — which converted an afternoon with no standing into an afternoon with standing.
That is a cheap intervention and it is not a trick. If an organisation genuinely wants the writing, saying so where it counts is the honest form of wanting it.
The specific wording matters
"Documentation" as a criterion produces documentation-shaped output: long pages, tidy structures, the twelve-page guide.
What works better is naming the thing the team actually needs. Some version of: did you write down things that other people needed, and did they use them?
That phrasing rewards the four-line answer that nine people opened and does not reward the glossary nobody read, which is the correct ordering and is not what a criterion called "documentation" produces.
The evidence problem
A criterion needs evidence, and the evidence for writing is weak for all the reasons in this section — the readings are invisible, nobody reports being helped.
Three workable sources:
- Authorship history, which most tools keep and nobody reads.
- Open counts, where available, with the caveats on the next page.
- The person's own list of what they wrote, which is the simplest and is normal for every other kind of work.
The third is enough. Nobody demands instrumented evidence for having shipped a feature.
The failure mode to watch for
Any criterion with a number attached produces the number.
A target of two pages a quarter produces two pages a quarter, of whatever length is cheapest, on whatever subject is easiest, read by nobody. This has its own page later in this section because it is common enough to deserve one.
The criterion should be qualitative and evidenced, not counted. "Here are three things I wrote and here is who used them" is a sentence that cannot be gamed cheaply.
Who this is available to
Only the person who runs the review, which is a narrow audience.
For everybody else, the version within reach is smaller: mention what you wrote when you describe your quarter. Most people do not, for the reasons in the second page of this collection, and the omission is self-inflicted.
If the writing appears nowhere in what somebody says about their own work, it will certainly appear nowhere in what anybody else says about it.
Wording the criterion
"Documentation" as a criterion produces documentation-shaped output. Naming the thing the team needs produces the thing the team needs.
- Produces guides nobody reads: "Contributes to team documentation."
- Produces pages people use: "Wrote down things other people needed, and they used them."
- Produces twelve pages a quarter: any version with a number in it.
The second phrasing rewards a four-line note that nine people opened and does not reward a glossary nobody read, which is the correct ordering. It also cannot be satisfied by volume, because the second clause requires somebody else to have turned up.
The honest limit
Naming writing in a review criterion does not make it rewarding. It makes it defensible, which is a lower and more achievable thing.
A person who spends Thursday writing can now explain the afternoon. That is the whole of what this intervention delivers, and on the evidence above it is enough to move three people from writing nothing to writing something.