The Recall

What this pack runs.

A defect has been found in some of what you sold. Some units are confirmed bad and traced to a lot; a larger number share the cause and cannot be told apart from them, because the cause is a supplier's component across many lots, a process every unit went through, or a source no one has found. The traced units are the only figure with a document behind them, so they become the recall by default, and the room prices the full take-back against them and finds it unaffordable. That is not the comparison. This pack turns it into two multiplications and a division. What taking them all back costs is every unit you can reach, at what reaching it costs from where it is, with the share of customers who will act on a notice stated rather than assumed. What leaving them out costs is the same units, times the share that are bad, times what one bad unit left out costs. The ratio of what one unit costs to take back to what one bad unit costs to leave is the break-even defect rate, how many of the untraced units must be bad before taking them all back is cheaper, and it turns a fight about the defect into a question the supplier's test can answer. Then the day: every day the room does not decide, units move from a warehouse where they are cheap to reach to a home where they are not, and a day adds to the cost of taking them back and nothing to the cost of leaving them out, so the model counts the days left before the default becomes the cheaper option. It stops on two halts: no one has drawn the line around the units that share the cause, and no one has put a price on one failure in the field.

The casebooks these cases come from.

The pack reads nine sourced decisions through the framework. They are drawn from these casebooks, each one a set of verified records on a single decision type.

A verified record has passed a mechanical claim-versus-source check, a second model's read, and human approval. The count is what exists today, not a target.