The review budget, and the rate at which checking stops working

The hour nobody put in a calendar

Six weeks produced 1,310.9 minutes of checking and repair that had not existed before: 21.85 hours, across four merchandisers, 54.6 minutes each a week.

Nobody scheduled it. It arrived in the gaps, between a call from the cutting room and a mill chasing a payment. It was done by people whose day was already full, because the saving it was paying for had been spent the moment it appeared.

Put the two totals side by side: for every minute of net saving, 0.92 minutes of checking appeared. That ratio is the honest shape of the change. Not an hour returned, but an hour of doing traded for fifty-five minutes of reviewing. And the trade is only worth making if the reviewing actually happens at a rate that works.

What "at a rate that works" means, measured

Marasem's log recorded when each check started and ended. So the desk could count how many checks a person did within the same hour. Grouping the six weeks that way produces the only figures in this course that genuinely surprised the people who generated them.

A defect counts as known if it was found at review, or found later by anyone: a colleague, a supplier, an inspector, an invoice. 47 were known in the six weeks. 31 were found at review.

Same four people. Same six weeks. Same tasks, distributed across the three rows by nothing more than what else was happening that hour.

The densest row held 14 of the 47 defects — 29.8% — and produced 8 of the 16 that got through, which is 50.0% of every escape in the log.

Why this is the failure mode rather than an unlucky week

The mechanism is a circle, and it closes on itself.

Handing work over creates time. The time is spent immediately, because a merchandising desk has no shortage of work to move into it. The checking that the handed-over work requires now has to fit into a day exactly as full as it was before. So checks get packed closer together. Packed closer together, they catch less. Catching less, more defects reach the buyer, the floor and the invoice. And dealing with those eats the time that would have paid for checking.

Marasem's fifth week is that circle in one data point. An order shipped that week, and the desk logged 51 of its 214 uses — 23.8% of the six weeks' work. That week produced 7 of the 16 escapes, which is 43.8%.

The busiest week did the most handing over and the least checking. That is precisely backwards, and it is what any desk will do unless the review time is protected in advance.

What the desk changed

Two rules, neither of them about the tool.

A ceiling on checks in an hour. Marasem set five. Beyond five, the sixth job either waits for the next morning or is done by hand. This is uncomfortable and it is the whole point: the honest response to no time to check is not to hand the work over.

The saving is booked before it is spent. Each merchandiser holds one protected hour a day for review. It is not a productivity hour, and it does not absorb overflow. It is the hour that makes the other saving real. If it is being eaten every day, the desk is running a deficit it cannot see.

Who carries it

One more thing belongs in a lesson about review, and it takes a sentence.

The merchandiser signs the cost sheet, the shipping instruction and the email. Nothing in the log changed that, and no arrangement of these tools changes it. A figure that leaves the desk is the desk's figure, whoever first typed it. That is why the review budget is not an administrative nicety. It is the only place where the responsibility the desk already carries is actually discharged. What a factory owes its buyers and its people when it works this way, and what a contract should say about it, is course 15.5's subject.