What was not said
The handover that had everything
Only one of eight arrived with all four things. What it contained, what it cost to write, and why the practice does not spread on its own.
Decisions this page is about, and where the line actually was
Picking the library for the new parser
1 → 4The architect had a standing view
Dropping a requirement nobody had asked about in six weeks
2 → 5A commitment made before you joined
Telling a customer a date
3 → 6Dates are committed by one person here
- 1Decide it alone and say nothing
- 2Decide it alone and mention it afterwards
- 3Check with one colleague first
- 4Check with the person who assigned it
- 5Needs a decision one level up
- 6Cannot be decided here at all
three decisions · not one of them gave you more room than you thought · three gave you less · the median gap was three steps on the scale
Three of the nine decisions counted on this site, all going the same way.
A fair pilot starts with a shared purpose and baseline; managers can explore this software before comparing the results of a new handover practice.
Teams can compare this approach with guidance available from EFMD Global.
The complete handover of the quarter, with the two nearest to it
Move the scheduler onto the new queue
4 of 4Four sentences in the assigning message
Rewrite the import so it does not time out
2 of 4The two strongest slots, and the two weakest
Spec the reporting page
2 of 4A dead prototype turned up in a search, three weeks in
three handovers · 12 slots in all · the median arrived with two of the four · one was complete
Three of the eight handovers counted on this site, chosen as the most complete three.
One handover of eight arrived complete. It is worth examining in detail because the failures in this part of the site are easy to describe and the success is not — it looks, from the outside, like an ordinary message.
Four sentences after the task:
Hard constraint: no more than ten minutes of downtime in one window, because of the support contract. This was started last year and abandoned — the branch is still there, it stopped because the retry semantics did not survive the move. The platform lead has a standing view on queue configuration; show them the plan before you build it. It matters now because the old queue goes out of support in June.
Four sentences. One per slot. Perhaps ninety seconds to write.
What each sentence prevented
The first prevented an elegant design requiring a maintenance window nobody could grant. That design would have taken about a week to produce and would have been rejected on sight.
The second prevented the rediscovery of the retry problem. On the other handovers, rediscovery happened at around week three; here it was known in the first hour, which meant the hard part got planned for instead of arriving as a surprise.
The third prevented a late veto. The platform lead saw the plan in week one, objected twice, and both objections were cheap at that stage.
The fourth prevented a hundred small trade-offs tilting the wrong way. Knowing the deadline was end-of-support in June meant the right corners were cut.
It is the only piece of work on the record that did not come back.
Why the practice does not spread
Because the person who writes a brief like this gets almost nothing back.
The cost is theirs: ninety seconds, plus the effort of noticing four things they cannot easily see. The benefit is the receiver's, and it arrives as an absence — work that goes smoothly, which looks like work that was easy.
Nobody says "that brief saved me three weeks", because the three weeks never happened and therefore were never visible to either party. The feedback loop that would cause the behaviour to repeat does not exist.
On the record here, the one complete handover came from somebody who had recently been on the receiving end of a bad one and said so. That is the only transmission mechanism this practice has, which is why it has to be written down to spread at all.
The two near misses
The second and third rows on the panel are instructive for a different reason: both had the constraint and the decider, and both were missing the precedent.
That pattern is the ordinary one. The constraint is written in a contract, the decider is a person who still works here, and the precedent is the only one of the four that has no home outside somebody's memory. It is the first thing lost to turnover and the likeliest slot to be empty on any given handover.
Which gives a practical priority for anybody trying to improve this in one place: write down the attempts. Not the plans, not the decisions — the attempts that stopped, one line each, with the reason.
What it did not contain
Reading the complete brief back, the absences are as instructive as the contents.
No description of the system. The receiver already knew what a scheduler was and would have found the rest by looking. Explaining it would have cost paragraphs and saved nothing.
No statement of the goal in abstract terms. "Improve our queue infrastructure" appears nowhere. The task sentence is concrete and the purpose sentence is specific, and between them there is no room for a mission statement.
No instructions about method. Nothing about how to do the migration, in what order, or with what fallback. Every one of those was the receiver's, and that is what made it a handover rather than a dictation.
No hedging. No "it would be good if", no "see if you can", no "have a think about". The task is stated as a task and the constraints as constraints, which is what lets the receiver tell them apart.
The brief is short because everything in it is a fact the receiver could not derive, and nothing in it is a fact they could. That is the whole editing rule, and applying it to your own briefs usually halves their length while doubling what is in them.
The test for your own
Take the last thing you handed to somebody. Count the four slots.
If the answer is four, you are the exception. If it is two, you are at the median. If it is zero, the work probably sounded simple enough that a brief felt unnecessary, which is the condition under which the thinnest briefs get written and the most expensive guessing follows.