Skip to content

Linear Issue Tracking

Home / Guides & Runbooks / Linear Issue Tracking

How we structure work in Linear — the hierarchy, what belongs at each level, and the conventions that keep it legible as the team and the codebase grow.

Hierarchy

Level What it is Timescale Example
Initiative A company milestone the whole team drives toward Quarter (~3 months) "Ship first fully-automated part end-to-end"
Project A bounded outcome serving an initiative Weeks to ~2 months "CAM v1 Toolpath Generation"
Issue A mid-sized step serving a project Days to ~1 week "Slot toolpath generation"
Sub-issue A single PR-level step serving an issue Hours to ~1 day "Slot detection"
Label A category that cuts across projects / initiatives Domains ("Fixturing", "Linking"), components ("Ruffles", "Hardware Assets")

Initiatives are milestones, not product areas

An initiative is a company milestone — a landmark outcome the org rallies behind on a month-to-quarter horizon, not a standing product area or a single dated feature. Frame it as what the company must achieve this quarter:

  • "Ship first fully-automated part end-to-end"
  • "AutoCAM running in production for internal jobs"
  • "First paying customer onboarded"

A product-scoped effort like toolpath generation is a project under one of those initiatives — not an initiative itself. Rule of thumb: if a candidate initiative names a component or a feature rather than an outcome, it's probably a project.

Practical Guidelines

A common workflow: track the work for a mid-sized deliverable as sub-issues under one issue, then work each sub-issue one at a time as an individual PR. This keeps a reviewable PR mapped to exactly one sub-issue.

Use views heavily

Whenever you want a clear slice of issues matching certain filters, create a View. Views can be team-level or private.

Examples:

  • Kanban: My Issues
  • Kanban: Issues with the label "Fixturing"
  • List: Issues under the "Ship first automated part" initiative

Use labels for cross-project / cross-initiative categories

Labels earn their place by cutting across the hierarchy. A good label:

  • Spans a meaningful number of issues (~10+)
  • Does not fully overlap an existing project

For example, "Benchmarking" would be a poor label because there's already a Benchmarking project — the label would carry no information the project doesn't.