SLOs have a reputation as a heavyweight ritual — quarterly reviews, spreadsheets, and a working group. They can be that. They can also be one number that quietly settles arguments for you.
Pick one thing users feel
A good SLO measures something a user would actually notice. For most request-driven services that is "did the request succeed, and was it fast enough?" Resist the urge to track everything. One or two indicators that map to real experience beat a dashboard of twenty that map to none.
Let the budget decide
The point of an error budget is to move decisions from opinion to data. When the budget is healthy, ship features. When it is spent, the same rule says stop and stabilize — no meeting required, because the number already made the call.
slo "checkout availability":
sli: rate(success) / rate(total)
target: 99.9% over 30d
budget_burn_alert: 2% in 1hAlert on burn rate, not on breach
- A fast burn — spending an hour's budget in minutes — is a page.
- A slow burn — trending toward breach over days — is a ticket.
- A healthy budget is silence, which is the whole point.
The only meeting you need
Once a month, glance at whether the target still matches reality. If you never miss it, the target is too loose. If you always miss it, it is too tight or the service needs work. That five-minute check is the entire process.