Teaching a new grader: the first week on the line

The grading line is the most consequential job on our yard. A grader who is 10 percent loose costs us customers; one who is 10 percent tight costs us inventory. Here is how we teach it.

Day one: the order matters more than the criteria

New graders learn the six points in a fixed sequence and are not allowed to vary it: corners, bottom seam, walls, flaps, moisture, odour. Always that order, every box, no exceptions.

The sequence is not arbitrary. Corners come first because they are the most consequential and the fastest to check — if the corners fail, nothing else matters and you have spent four seconds. People who improvise their own order start missing things within a week, because they check the interesting parts and skip the boring ones.

Day one, second half: the reference set

We keep a shelf of about twenty boxes that never get sold. Two clear Grade A. Two clear B. Two clear C. Two obvious scrap. And a dozen deliberately awkward ones: a beautiful box with a soft corner, an ugly box that is structurally perfect, one with a tide line you have to look for, one that smells faintly of solvent.

New graders work the reference set until they get all twenty right twice in a row. It usually takes two or three sessions. The awkward twelve are where the actual learning happens.

The two mistakes everybody makes

1. Grading on appearance

A clean-looking box feels like an A. A grubby one feels like a C. Neither feeling is data. The soft-corner box on the reference shelf is beautiful and it is scrap; the ink-stained one is ugly and it will do three more trips.

This corrects itself within a week if the reference set includes enough counter-examples. It does not correct itself at all if people learn on real inventory, because real inventory is mostly correlated — grubby boxes usually are worse — and the correlation teaches the wrong lesson.

2. Grading on sympathy

This one is subtler and takes longer. A grader who knows the load came from a customer who is having a hard year starts finding reasons to call boxes B instead of C. It is a decent human impulse and it is corrosive, because the customer's settlement is now unpredictable and their next load will be graded by someone else.

Our rule: grade the box, not the situation. If there is a commercial reason to be generous, that decision is made on the settlement, in the open, by someone whose job it is — not quietly on the line.

Week one: the double-grade

For the first two weeks, a second grader re-grades a random 10 percent of everything a new person produces. Disagreements are resolved at the box, not in a meeting, and both people say what they saw before anyone announces an answer.

It is slow and it is worth it. We know within days whether someone runs tight or loose, and either is fixable if you find it early.

Ongoing: monthly calibration

Everyone, including the people who have been doing it for a decade, re-runs the reference set once a month. Drift is real. It is usually gradual and in one direction, and nobody notices it in themselves.

The part that makes it stick

New graders spend a day at the receiving end: opening a load a customer would have received and seeing what it feels like from that side. It changes how people grade more than any amount of instruction.

The same is true in the other direction. Everyone on our quote desk spends a month on the line before they write quotes, because you cannot describe a Grade B box you have never held.