The lag
Why a problem arrives through a side door
Not one of the five costly removals was reported by anybody inside the process it belonged to.
The standing slot for work with no ownerThe fastest of the five problems
week 5 · surfaced after 14 weeks
Reading supplier contracts before renewal
week 2 · surfaced after 22 weeks
The quarterly reconciliation of the project ledger
week 3 · surfaced after 31 weeks
Shadowing new starters through their first week
week 6 · surfaced after 47 weeks
Logging why a job was turned downThe slowest on the record
week 9 · surfaced after 64 weeks
Surfaced 5
Five removals were written down with the week each of them stopped. Every one of them produced an event inside the 78 weeks watched. Of these, five surfaced as a problem. All five removals whose absence surfaced as a problem, in the order they surfaced.
One detail in the record of the five costly removals is easy to pass over and hard to explain away. Not one of them was reported by a person doing the work.
Every one arrived through something outside: an audit, an invoice, a backlog count, a resignation pattern, a repeated request. In each case the organisation learned about the removal from a channel that had nothing to do with the removal.
Compare the nineteen removals that were asked back or crept back. Every one of those was handled inside the process it belonged to, by somebody who did the work, usually within a fortnight.
Two completely different feedback systems
This looks like one spectrum — fast reporting at one end, slow at the other — and it is not. It is two mechanisms that do not overlap.
The internal channel runs on wanting. Somebody performs a task, reaches for a thing that is no longer there, and either asks for it or rebuilds it themselves. The signal is a person's own inconvenience, which is immediate, reliable and cheap to act on.
The external channel runs on consequence. Something happens elsewhere in the world — a date arrives, a person leaves, a figure is requested — and the absence becomes visible from outside as a shortfall. The signal is somebody else's problem, which arrives at the pace of the outside event.
A removal is served by one or the other, almost never both. Whether it is served by the first depends on a single thing: whether anybody's own daily work is made worse by the absence.
Why the dangerous ones have no internal channel
The five costly removals all have the same structure: the people doing the work were not inconvenienced.
- Nobody missed the standing slot, because the jobs that used to be picked up in it belonged to nobody by definition.
- Nobody missed the contract reading, because the day the contract renewed was not a day anybody was reading it.
- Nobody missed the reconciliation, because the quarter it covered was over.
- Nobody missed the shadowing, because the person who would have benefited had not arrived yet.
- Nobody missed the refusal log, because each person refusing a job found the reasoning themselves and did not know anybody else had.
In every case the person who could report is either absent, in the future, or unaware there is anything to report. That is not a failure of attention. There is nothing for attention to catch.
A procedure whose absence inconveniences nobody today has no internal channel, and the absence of complaints about it is not information.
The misreading this produces
An organisation watching its transition sees the internal channel working beautifully. Requests come in, get handled, and tail off. By week fourteen the stream has stopped.
The natural reading of a stream that has stopped is that there is nothing left to report. The correct reading is that the mechanism with a two-week response time has finished, and the mechanism with a fifty-week response time has not started.
Both mechanisms are present from day one. They simply have response times that differ by a factor of twenty, and only one of them produces a visible stream.
What this implies for watching
If the internal channel cannot report on a category of removal, then watching harder does not help, and neither does asking people whether anything is missing. The honest answer from everybody doing the work will be no, and that answer is correct about their work.
What is left is narrower and more mechanical:
- For each removal, name the external event that would make its absence visible — the renewal, the quarter end, the induction, the audit.
- Put the next occurrence of that event in a calendar.
- On that date, ask the specific question rather than the general one. Not "is anything missing?" but "when this renewal came round, did anyone read the terms?"
The general question routes to the internal channel, which has nothing to say. The specific question stands the asker, briefly, in the place where the absence is visible.
On this record that would have cost perhaps four questions a quarter and would have caught three of the five removals a great deal earlier than they were caught.
The one that nearly had an internal channel
The reconciliation is worth isolating, because it came closer than the other four to reporting itself and the reason it did not is specific.
At the first quarter end after it stopped, somebody in the team noticed that the quarterly figures had not been assembled. They mentioned it to nobody, because the figures were not needed by anybody that quarter and assembling them would have been work for no purpose.
That is a correct judgement. It is also the moment at which the internal channel had something to say and declined to say it, on the reasonable grounds that there was nothing wrong.
The thing that was wrong arrived eight months later, when four quarters had to be assembled at once under time pressure. The cost was not in any single quarter's figures; it was in the accumulation, and accumulation is invisible at every point until it is large.
A procedure whose failure mode is accumulation has an internal channel that works perfectly and reports nothing, because at each individual point there is genuinely nothing to report.
Also on what stopped and when anyone noticed