What came back
What came back, and at what weight
Twenty-four of the thirty-eight returned in some form, at a median of just under half the load they carried before.
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.
Reading supplier contracts before renewalReinstated in full after the second unread renewal
100%
Checking the diary a day ahead
100%
The daily stand-upTwice a week, standing, and timed
30%
Visits to the two furthest sitesFour a year instead of monthly
30%
The Monday planning hour
45%
Two people on every delivery
70%
The spare key log
20%
Keeping the stock list on paper as well
25%
Shadowing new starters through their first weekA named person to ask, instead of a week alongside
50%
Same 2Reduced 4Substitute 3
Nine removals came back in some form, at a median of 45 per cent of what the same thing cost before. Two of them came back whole; the other seven came back smaller or came back as something else. One removal is not shown here at all, because it never came back and asking what it came back as would not be a question about it. One is not shown because it was marked as the same thing at a different load, which is a smaller version by definition. Nine of the twenty-four, with two lines the panel refuses.
Twenty-four of the thirty-eight came back in some form. Put like that it sounds as though the shorter week recovered two thirds of what it had removed and the real saving was fourteen things out of thirty-eight.
That reading is wrong, and the record says so clearly enough that a site called Nobody Missed It Yet is obliged to print the correction on its own front page.
Three ways of coming back
The same thing, at the same load. Six of the twenty-four. The procedure was reinstated as it had been, with the same frequency and the same people. Five of the six followed an incident rather than a request — something went wrong and the old arrangement went back in.
A smaller version. Twelve. Fewer people, less often, shorter, or with a step taken out. The daily stand-up came back twice a week, standing and timed. Monthly site visits came back as four a year. The Monday planning hour came back at under half its length.
Something else doing the same job. Six. The need turned out to be real and the old mechanism turned out not to be the only way of meeting it. The paper stock list was replaced by a weekly export at a quarter of the effort. The spare key log became a labelled board.
The number
Against the load each thing had carried before, the median return was forty-seven and a half per cent.
So the twenty-four that came back did not undo the subtraction. They came back at under half their old weight, and the fourteen that did not come back at all are a permanent reduction.
The honest summary is not "fourteen out of thirty-eight". It is something harder to put on a slide: a third of the list gone entirely, and most of the rest on a diet.
Why this is the strongest number here
Everything else on this site is a warning. This is the one figure that runs the other way, and it runs further than the headline does.
If an organisation removes thirty-eight procedures and ends up carrying fourteen fewer plus twenty-four at roughly half weight, the arithmetic of hours recovered is substantially better than the count of removals suggests. A reinstatement is not a reversal. On this record it was, in three cases out of four, a renegotiation that landed somewhere neither the original procedure nor its removal had been.
This is also why the usual framing of a trial — how many things did you have to put back? — produces a misleading answer. Twenty-four sounds like failure. Twenty-four at a median of forty-seven and a half per cent sounds like what it is.
Why it also makes the warning worse
Cheap returns and expensive discoveries are the same fact seen from two ends, and the end you see depends on whether anybody was watching.
What comes back goes through a filter. Somebody has to want it enough to ask or to restart it, and what they restart is the part they actually wanted — hence the half weight. The filter is the wanting, and the wanting is what makes it cheap.
What surfaces as a problem goes through no filter at all. Nobody wanted anything; an event occurred. It arrives at full price, through an audit or a resignation or an unread contract, and it arrives late. Of the five problems on this record, one was reinstated at full weight, two came back as substitutes, and two produced costs that were simply absorbed.
Everything that came back through somebody wanting it came back smaller. Everything that came back through something going wrong came back at whatever size the problem demanded.
What the panel refuses
Two lines above are in the data and absent from the picture.
One is marked as the same thing at a different load. That cannot be true in the sense the column means: the same thing at a different load is a smaller version, which is a different row. The contradiction is in the record rather than in the thing, and the panel says so rather than quietly picking one of the two readings.
The other never came back at all. Asking what it came back as is not a question about it — it is a question with no subject — and a blank bar would have read as a return of size zero, which is a different claim from no return. The two are confused constantly, and the confusion is one of the things this site is about.
What a return costs to arrange
The weights above describe what came back. They do not describe what it took to get it back, and the two are different enough to be worth separating.
A request costs a conversation. Somebody asks, somebody else explains that the hours are not there, and a version emerges. On this record that conversation happened eight times and none of them took more than a few minutes.
A quiet return costs nothing visible, because nobody arranges it. It assembles itself out of individual habit, and the organisation does not find out until somebody asks.
An incident-driven return costs whatever the incident cost, plus the restoration. All five of the problems on this record consumed management attention out of proportion to the thing being restored — the reconciliation took a fortnight of unplanned work before anybody got to the question of what should go back.
So the three channels differ by roughly two orders of magnitude in arrangement cost, and the weight of what comes back runs in the opposite direction to the cost of getting it there.