Economics Before Metrics

Start by pricing the error, not by measuring the metric.

What does a miss actually cost? Not in service points. In the thing the organization loses: a stranded repair job, a technician idle for a day, a customer who buys the next one somewhere else, cash tied up that had another use.

The moment those numbers exist, most rankings become obvious, and a good number of popular initiatives stop being defensible.

Metrics that pretend to be constraints

“We must hold 95 percent fill rate” sounds like a constraint. It is usually a preference with a history.

The test is to ask what happens at 94 percent. If the answer is a specific, named consequence, it is a constraint. If the answer is that someone will be unhappy, or that it is the number in the deck, it is a target that has been promoted above its evidence, and treating it as inviolable will rank your alternatives for you before any analysis runs.

This matters because a false constraint does not merely distort the answer. It hides the trade. An organization that will not discuss 94 percent cannot discover that the last point of service costs four times the previous one, which is frequently the most valuable finding available.

When a technical metric earns its keep

Technical metrics are not banned. They earn their place when they are a faithful proxy for a consequence and the proxy relationship has been checked.

Forecast accuracy earns its keep when a named improvement changes an order quantity enough to change a cost. It does not earn its keep as a standalone objective, because most accuracy improvements do not cross any decision threshold. A model that improves mean absolute percentage error by three points and changes no order is a diagnostic with a press release.

When the units will not reduce

Sometimes two consequences genuinely will not reduce to one number: stranded jobs against working capital, service against cash.

Do not force a conversion by inventing a dollar value for something nobody has priced. Do the opposite: state the exchange rate explicitly as an assumption and show where the ranking flips as it varies.

“We recommend class-based stocking if a stranded job is worth more than roughly 340 dollars of tied-up capital. Below that, the blanket cut wins.” That sentence hands the trade to the person whose job it is to make it, with the quantitative structure attached. It is more useful than a single blended score, and it is considerably more honest.

Dollars, not because dollars are holy

The argument for money as the default unit is not that money is the point. It is that money is the only unit in which an organization’s incommensurable concerns have already been forced into comparison, every day, by everyone.

Where dollars are dishonest, use the unit that is not. Stranded jobs, hours of downtime, missed clinical appointments. The requirement is that the unit bites, meaning that a person in the room feels the number rather than files it.