error-budgets
Error Budgets
Every organization has an implicit, unstated tradeoff between shipping fast and staying reliable — it just usually gets negotiated informally, incident by incident, by whoever's loudest in the room. An error budget makes that tradeoff explicit and quantitative: given your SLO, you're allowed a specific, calculated amount of unreliability, and you get to decide how to spend it — on risky launches, on infrastructure migrations, on just moving fast — instead of debating each decision from scratch.
The budget only does its job if spending it has a real consequence. A budget that's tracked but never enforced is a dashboard, not a policy — it changes nobody's behavior and reliability erodes exactly like it would without one.
The budget is a spendable number, not a report card — and it must have teeth.