Which broken thing to fix first

Warranty claims, product defects, service failures, technical remediation, they share a shape: more things need fixing than there is capacity to fix them, and the cost of getting the order wrong is large and mostly invisible. The usual triage is severity-first or loudest-first. Both ignore what matters: the downstream cost of leaving each case unaddressed.

The decision

Given a limited remedial budget, people, parts, time, allocate it to maximise avoided loss, subject to constraints that cannot be broken: contractual deadlines, safety floors, customer commitments. That means estimating, for each case, the expected downstream loss if it waits, degradation, escalation, contractual penalty, churn, reputational exposure — and then treating the assignment of effort as a constrained allocation problem, not a sorted list. When cases compete for the same scarce skill or part, this is naturally a min-cost-flow or assignment structure rather than a spreadsheet sort.

Repair, replace, compensate or defer

The same frame answers the neighbouring question. For each case there is rarely one action but several, fix now, replace, compensate the customer, or defer, each with a cost and an effect on downstream loss. The right move is the one that buys the most avoided loss per unit of scarce capacity, and that is frequently not the most thorough fix.

What this is

This is a method for ordering effort, labelled as an idea. It does not manufacture capacity, and it is only as good as the loss and exposure estimates it is given, where those are weak, the output is a structured guess, and the deliverable calls it one. What it changes is that the guess becomes explicit, comparable and improvable, instead of buried in whoever shouted last.