The four kinds of thing that stopped

The four groups

Error-catching — eleven entries. Second reviewers, duplicate records, counts, reconciliations, the second quote on small purchases. These exist so that a mistake is found before it costs anything.

Knowledge-carrying — nine entries. Handover notes, the induction folder, shadowing new starters, the log of refused jobs, the written note after a phone order. These exist so that what one person knows reaches another person.

Attention-keeping — twelve entries. Standing slots, weekly circulations, reminders, the wall planner, the diary check, the monthly stock count. These exist so that something does not drift out of view.

Occasion-marking — six entries. The away day, Thursday biscuits, the suggestion box, the annual laminating of the safety notices. These exist for reasons that are mostly not operational.

How each group behaves when removed

Read the rows rather than the totals. Four rows of nine to twelve entries cannot establish a rate. What they can show is a shape, and the shapes differ in ways that are explicable.

Attention-keeping things creep

Six of the eleven removals that crept back were attention-keeping, and that is the largest single concentration anywhere in the table.

The mechanism is almost mechanical. If a procedure exists to stop something drifting out of view, then removing it lets the thing drift, and the drift is noticed by whoever cares about the thing. They do not raise it. They simply start checking again, informally, because it is easier than arguing. The procedure reassembles itself out of individual habit, usually in a smaller form.

This is the benign end of the record. Nothing is lost and the cost of the thing is reduced, because what reassembles is only the part somebody actually wanted. The stand-up came back twice a week. The stock count came back at forty per cent of its old load.

Occasion-marking things do not come back at all

Four of the six were never missed, and the two that were asked back were asked back fast — three weeks and one week.

That is a clean split and it is the most predictable group on the list. An occasion is either wanted, in which case somebody says so almost immediately because the next occasion is due, or it is not, in which case nothing is waiting to remind anybody of it.

There is no slow mode here. If an away day is going to be missed it is missed at the time the away day would have been.

Knowledge-carrying things do not signal at all

Four of the five costly removals on this record came from this group of nine, and not one of the nine was asked back.

Nobody asks for the return of a procedure that moves knowledge between people, because the person who would ask is the person who does not have the knowledge, and they do not know they are missing it. The absence announces itself to a third party, much later, in some other form: an auditor who asks for figures, a new starter who leaves, a contract signed on terms nobody read.

An error-catching procedure fails by letting an error through. A knowledge-carrying one fails by producing a person who does not know something, and nothing about that person looks different.

What to do with this

The table does not say keep the knowledge-carrying ones. Three of the nine were never missed, including two logs that on inspection had never been read back by anybody.

It says something narrower and more usable: the group tells you what kind of feedback to expect, and therefore whether waiting for feedback is a plan.

• For attention-keeping removals, waiting is a plan. You will see it come back, and what comes back is cheaper than what went.
• For occasion-marking removals, waiting is a plan, and the wait is one cycle long.
• For error-catching removals, waiting is a weak plan. You will probably find out, but possibly only when the error is expensive.
• For knowledge-carrying removals, waiting is not a plan at all, and the only way to find out is to go and ask somebody who would know.

That last line is the whole practical content of this page. It applies to nine items out of thirty-eight, and on this record it would have been worth about an hour a quarter.

Where the two ambiguous entries went

Two entries could plausibly have gone into a second group, and both are noted here rather than quietly assigned.

The written note after a phone order is error-catching if you think its job is to stop orders being lost, and knowledge-carrying if you think its job is to tell the next person what was agreed. It was counted as knowledge-carrying on the strength of what it actually contained, which was mostly context rather than confirmation.

The month-end figures circulated to everyone is attention-keeping if the point is that people see the numbers, and occasion-marking if the point is the ritual of the month ending. It was counted as attention-keeping.

Moving either one changes the table by a single entry and changes none of the conclusions, which is worth saying explicitly: a sorting that the result depends on is a sorting that should not be trusted.