Error Budget

An error budget is the amount of unreliability you're allowed to spend in a given period — the flip side of a service level objective. If your SLO says the API should succeed 99.9% of the time this month, the remaining 0.1% is your error budget: a concrete allowance of failed requests or downtime. The idea reframes reliability as a resource to spend deliberately rather than a vague goal of "no outages." When the budget has room, you can ship faster and take risks; when it's nearly exhausted, the team slows down and prioritizes stability. For founders, this turns the endless velocity-vs-reliability argument into a data-driven policy that product and engineering agree on in advance. Practical note: set the SLO from what users actually need, not 100% — perfect reliability is infinitely expensive and blocks all shipping. Agree up front on what happens when the budget runs out, so the rule is automatic instead of a fight during an incident.