What metrics should I start with if I can only do two?

Start with PR cycle time (opened → merged) and CI queue time vs runtime. Those two expose review and infra bottlenecks that usually dominate lead time.

How do I avoid surveilling engineers?

Aggregate at team/repo level, not per person. Use timestamps from Git and CI, not IDE plugins or activity trackers. Share the dashboard openly and focus on system fixes.

Do I need Backstage to do this?

No. A template repo + reusable CI + a README can deliver 80% of the value. Add Backstage later to advertise golden paths once they’re stable.

What about monorepos vs polyrepos?

The playbook works for both. In monorepos, be extra disciplined about changed-file test selection, concurrency cancellation, and preview env routing by path.

Will AI code assistants fix this?

AI speeds up local coding, not organizational queues. If PRs sit for a day and CI queues for 20 minutes, AI won’t help until you kill those waits.

Platform-productivity · Oct 3, 2025 · 9 minute read

The Hidden Queue: Measuring Dev Friction and Killing Hand‑Off Wait Time on the Paved Road

If your PRs idle for a day and CI sits in “queued” for 20 minutes, you don’t have a talent problem—you have a queueing problem. Measure the friction that matters and pave the road so devs never wait on humans or bespoke tools again.

Back to all posts

The Hidden Queue: Measuring Dev Friction and Killing Hand‑Off Wait Time on the Paved Road

Key takeaways

Implementation checklist