Skip to lesson
Exit
Bug Triage & Prioritization at Scale1 / 3

1 min lesson

Severity: how bad is the failure itself

Connect "How bad is the failure itself" to the decision you would make.

Step 1 of 3

Severity: how bad is the failure itselfone report, judged on its worst outcome

Severity describes the damage independent of how many people hit it. A single user losing uncommitted work is a serious event even if they're the only one. Rank failure classes from the one you'd interrupt someone over down to the one that can wait a sprint.

Learn more

Advanced table

Severity is graded on the worst plausible outcome, not the average one

Tier
S0 - Critical
What it means
Data loss, security exposure or full outage
Cursor example
An Agent edit silently overwrites unsaved files; a privacy-mode leak; backend down for all users
Tier
S1 - Major
What it means
Core workflow blocked, no workaround
Cursor example
Tab autocomplete returns nothing for everyone on the latest update; Agent loops and never completes
Tier
S2 - Degraded
What it means
Works but worse or a workaround exists
Cursor example
latency doubled; codebase indexing is slow but eventually completes
Tier
S3 - Cosmetic
What it means
Annoyance, no functional loss
Cursor example
A misaligned icon in the model picker; a typo in a tooltip

Severity is graded on the worst plausible outcome, not the average one.

Watch out

Anything that can destroy a user's code or expose their data is S0 even if exactly one person reported it. Data loss has no quorum. If you find yourself debating whether a single-user data-loss bug is “really” urgent, you've already mis-graded it.