What came back
Six that came back whole
Six removals were reinstated exactly as they had been. Five followed an incident rather than a request, and that is why they came back at full weight.
The daily stand-upBack twice a week instead of five, standing and timed
week 0 · crept back after 3 weeks
Chasing unpaid invoices every weekBack in full
week 3 · crept back after 2 weeks
The spare key logReplaced by something else doing the same job
week 10 · crept back after 14 weeks
Two people on every delivery
week 4 · crept back after 9 weeks
Crept back 4
Four 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, four crept back. Four of the eleven that crept back without anyone deciding.
Six of the twenty-four returns came back at full weight: the same procedure, the same frequency, the same people. They are the smallest of the three groups and the most informative, because of what they have in common.
The six
- Reading supplier contracts before renewal.
- Printed copies of the weekly rota.
- The written handover at the end of a shift.
- Tea and biscuits on Thursdays.
- Checking the diary a day ahead.
- Chasing unpaid invoices every week.
Two of those are trivial and four are not, which is the first thing worth noticing: full restoration is not a property of important things. Thursday biscuits came back whole. So did a contract review that had cost the organisation an unread renewal.
What five of them share
Five of the six came back after something happened rather than after somebody asked in the abstract.
| removal | the event that restored it |
|---|---|
| contracts | an invoice arrived on terms nobody had read |
| printed rotas | somebody could not find the rota |
| shift handover | a job was started twice |
| diary check | a visit was missed |
| invoice chasing | the person simply never stopped |
The sixth, the biscuits, came back because somebody asked, three weeks in, and nobody had a reason to say no.
Why an incident restores at full weight
This is the mechanism, and it explains the whole group.
When a removal is reversed by a request, there is a conversation. Somebody says they want the thing; somebody else points out that the hours are not there; and what emerges is a version of the thing that fits the hours. That is how twelve entries on this record came back at a median of forty per cent.
When a removal is reversed by an incident, there is no conversation. Something has gone wrong, the thing that would have prevented it is known, and putting it back exactly as it was is the fastest available action and the one nobody can argue with. Negotiating the size of a precaution immediately after it has failed is a conversation nobody wants to have.
An incident does not restore the procedure that was needed. It restores the procedure that was removed, which is not the same thing and is usually larger.
The cost of that
On this record, at least two of the five incident-driven restorations look oversized in retrospect.
The contract reading went back as a full annual review of every supplier agreement. The two that caused the problem were the two whose terms had changed; the other eleven had not changed in years. A version covering changed terms only would have cost a fraction and caught both.
The shift handover went back as a written note at every shift end. The job that had been started twice was a particular kind of job, and the handover note for the other kind had never been read by anybody.
Neither of these is an error anybody should be blamed for. Both are the predictable result of restoring under pressure, at a moment when the argument for a smaller version sounds exactly like the argument that just failed.
What follows from it
If a removal is going to be reversed at all, it is cheaper to reverse it by request than by incident, and the difference on this record is roughly a factor of two in the weight of what comes back.
That is an argument for finding out early, and it is the practical payoff of everything else on this site. The value of a short lag is not only that the problem is smaller. It is that the thing which goes back in is smaller, because it goes back in through a conversation rather than through a reflex.
A note on the sixth
The biscuits are worth a line. They came back after three weeks, at full weight, because somebody asked and the cost was trivial. Nobody could name an operational effect and nobody pretended to.
On a record devoted to the invisible costs of removals, there is a category that is simply about whether a place is pleasant to work in, and it does not need to be justified in terms of output. Removing it cost nothing measurable. Putting it back cost nothing measurable. The only honest thing to say about it is that somebody missed it, which is a sufficient reason, and that it is the only entry on the list where the record contains no figure of any kind.
What a restoration does not restore
There is a second thing lost when a procedure is removed and later put back at full weight, and nobody on this record noticed it until the read-back.
The procedure comes back. The knowledge of why it is there does not.
Before the transition, the contract review existed because somebody had introduced it after a bad renewal in the past. After the transition it exists because of a different bad renewal, more recently, and the new reason has replaced the old one in everybody's account of it.
That sounds like an improvement — a fresher justification — and in one respect it is. But the old reason covered a case the new one does not. The original review had been introduced after a supplier quietly changed a liability clause; the recent failure was about pricing. The restored review is explicitly a pricing check.
A procedure that has been removed and restored carries the shape of its most recent failure, not the sum of the reasons it ever had. On this record that is a narrowing in at least two cases, and it is invisible, because the restored version looks like the original and has the same name.