Recovery: the number that is not on the measurement sheet

Two complaints, eleven weeks apart, from two departments

4 September. Two things land at Quillan on the same morning.

The merchandising team has a store report from three shops. The ecru five-packs on the shelf look warmer, almost cream, next to the ones that have been there since July. Same style, same supplier, same colour code.

The customer-service team has a different report. Returns on HS-244 have gone from 0.9% of packs sold to 3.6%, and the written reason on nearly all of them says a version of the same thing: they fall down.

Two departments, two problems, and no obvious connection. Nobody is wrong about anything they have seen.

What the measurement sheet says

The first thing anyone does is measure. Tanyeri's kept samples from both deliveries are pulled and measured. The welt is the ribbed band at the top of the sock, the part that holds it up. In both deliveries it is identical.

Everything is on specification. Everything has always been on specification. And one of these two deliveries falls down.

Course 1.2 teaches what growth is and how it is measured on a fabric. Growth is the stretch a fabric keeps instead of giving back. You stretch it to a set extension, hold it, release it, and measure it again. In hosiery the difference is that you cannot cut a strip out of a welt and test it flat. A welt is a tube. Its rib is under tension from the moment it is knitted, and the elastane in it is usually laid in rather than knitted: carried along the course and trapped by the cotton, so it is not even visible on the face of the fabric. The test has to be done on the whole sock, on a former, at a stated extension.

Quillan's specification does have that test. It says welt growth not more than 8.0% after thirty minutes. Nobody had run it since the sealing sample in March, because the measurement sheet kept passing.

One cause

Go back to 19 June, and to the last part of lesson 4.

The first delivery failed Quillan's after-wash length requirement: 6.5% against a 3.0% tolerance. The forms that would have fixed it properly were three weeks away. So the mill raised the boarding steam from 118 to 132 degrees Celsius and the dwell from 9 to 22 seconds. The after-wash length loss fell to 2.8%, and the technologist signed it off.

Boarding is not a length operation. It is a heat treatment applied to every fibre in the sock at once, and there were two other fibres in the sock.

The elastane in the welt was set harder than it had ever been set. Setting elastane relaxes the stress in it and permanently weakens how strongly it pulls back. A hotter, longer set gives a welt that measures the same and comes back less. Growth went from 4.2% to 11.6%, and the sock stays up for five washes and then stops.

The optical brightener in the ecru went the other way. Brighteners are organic molecules, and heat breaks them down. The ecru left the boarder slightly yellower. Nobody could see that against a sealed standard on the day, and everybody can see it on a shelf next to the June delivery.

Two complaints. One cause. And the cause was a correct decision, taken under a real constraint, signed off by the person whose job it is to sign it off, to fix a genuine failure the buyer was right to raise.

What it cost, and what it would have cost to catch

Quillan claims 6% against the second delivery: 3,600 dozen pairs at USD 11.40 is USD 2,462. The order's real margin, after the size-mix miss in lesson 2, was USD 5,358. So the claim is 46.0% of what was left.

The test that separates the two deliveries costs about USD 22 at Tanyeri's own laboratory and takes half an hour. Three a week, one per boarding shift, is USD 66. Between the boarding change on 19 June and the first complaint on 4 September there are 77 days, or 11.0 weeks. That is USD 726 of testing against a USD 2,462 claim. And the claim is the small part of it, because the returns and the marked-down ecru sit with the customer.

Act as a senior hosiery quality manager who has seen a boarding change come back as a customer complaint. A measurement problem on my order has been solved by changing a heat process, and I want the consequences found now rather than in September. Order facts: buyer [BUYER], PO [NUMBER], style [STYLE], fibre composition [LIST WITH PERCENTAGES], colours [LIST, AND SAY WHICH ARE UNDYED, PALE OR OPTICALLY BRIGHTENED], quantity [QTY], deliveries and their dates [LIST]. The problem that was fixed: [WHAT FAILED, THE MEASURED FIGURE, THE TOLERANCE]. The change made: [PROCESS, OLD SETTING, NEW SETTING, DATE IT WAS MADE, WHICH BATCHES ARE BEFORE AND AFTER]. The buyer's specification for anything other than measurement: [RECOVERY OR GROWTH LIMIT, SHADE TOLERANCE, HAND, FASTNESS]. Do the following. First, list every fibre and every component in the article that this heat treatment also touched, and say what the change plausibly does to each one, in the direction it does it. Second, tell me which of those consequences the buyer's measurement sheet CANNOT see, and why, being specific about relaxed measurements against properties under extension. Third, give me the tests that would separate a batch made before the change from one made after, with the cost and turnaround of each, and tell me the minimum number of samples that would be convincing. Fourth, tell me how long the feedback loop is for each consequence in the field, in washes or in wears, and therefore how much production would be made before a complaint could physically arrive. Fifth, put a number on the exposure: how much of the order was made after the change, what a discount claim on it would be at [PERCENT], and what that is against the order's margin. Sixth, tell me what should have been recorded on the batch card so that a complaint could be traced to a date, and write the change-control note I should have written. Do not tell me to add material to fix a property that a process destroyed until you have ruled out reversing the process.