Post

I think there's another take on error budgets that's much more relatable to any sized company. I see it as a gauge to balance risk with speed. On one end you moved too fast and broke things. On the other, governments move faster and you were probably too careful. It's useful as a feedback mechanism.

The error budget itself represents the acceptable risk. Even if you have "poor" practices, if things are within that risk tolerance, there's no need to dedicate resources to being more resilient. You can keep focused on features or gaining customers.

Violate the budget and you shift focus to reliability work and reducing risk. You make deployments better and you fix the things that burned the error budget. To me, it's all about having a metric that objectively tells you if you're being too risky or how much risk you can tolerate.