What is Triage?
Triage is the rapid assessment and categorization of incoming work items. Borrowed from medicine (where triage sorts patients by urgency), product triage sorts bugs, feature requests, support escalations, and stakeholder asks into priority buckets that determine when and how they are addressed.
Effective triage prevents two problems: critical issues getting buried in a backlog, and the team being constantly interrupted by non-critical issues that could wait.
Why Triage Matters
Without triage, teams use one of two failing strategies: they address issues in the order received (so a critical security bug waits behind 15 minor UI complaints), or they address whatever the loudest stakeholder demands (so strategy gives way to politics).
Triage creates a fair, transparent process for evaluating incoming work. It ensures critical issues get fast attention and non-critical items get appropriate scheduling.
How to Run Triage
Set severity levels with clear definitions. Common levels:
- P0/Critical: Production down, data loss, security breach. Drop everything.
- P1/High: Major feature broken for many users. Fix within 24 hours.
- P2/Medium: Significant bug with workaround. Schedule for next sprint.
- P3/Low: Minor issue or enhancement. Add to backlog.
Run a short daily triage meeting (15-20 minutes). Review new items, assign severity, assign an owner, and move on. Do not solve issues in triage; just classify and route them.
Create clear routing rules. P0s go directly to an on-call engineer. P1s get added to the current sprint. P2s go into the next sprint's candidates. P3s go to the backlog.
Track triage volume and patterns. If a specific feature area generates consistent P1 bugs, that signals a need for structural investment, not just bug fixes.
Triage in Practice
Linear uses an "Inbox" concept for triage. New issues land in an unsorted inbox. During triage, the PM moves each issue to the appropriate project and priority level. Items that are not actionable get archived immediately.
At Shopify, triage includes a "customer impact" assessment. Each triaged item is tagged with the number of affected merchants and the revenue impact. This data-driven triage prevents subjective priority inflation.
Common Pitfalls
- Everything is P0. If every bug is critical, nothing is. Enforce severity definitions strictly.
- Triage as a meeting, not a process. Triage should include follow-up: verify that assigned items are actually being addressed.
- PM-only triage. Engineering input is essential for assessing technical severity and effort. Include a tech lead.
- No escalation path. When a P3 bug affects a major customer, there should be a way to escalate it outside the normal triage process.
Related Concepts
Triage connects to incident management for production issues and the product backlog for feature requests. It is an operational practice within product ops. Effective triage requires clear prioritization criteria and strong stakeholder management.