What an error budget is for, after a bad week

An error budget is a boring object until a Monday morning when customers have already waited. Then everyone asks whether the week “used it up”. In sittings we almost never find a ledger that was updated as the pages arrived. We find a slide from last quarter and a pile of tickets.

The useful question is smaller. Which pages, with timestamps, spent the budget. Which incidents never opened the ledger because nothing paged. Application health monitoring insights of that kind sit in the alert history and in the night notes, not in the slide.

If the budget is expressed as a percentage nobody in the operations room can point to on a ticket, it is decoration. During an Application Health Review we ask the on-call lead to pick one bad week and walk it without the slide. The sentences that come out of that walk belong in the brief.

We do not set budgets for you. We write down whether the ones you already claim would have moved if last month’s incidents had been entered honestly.