SPACE and Beyond - Measuring Developer Productivity
summary
Stripe instruments developer workflow with a central Bad Day server plus Honeycomb and a data warehouse, emitting ~1.5M events/week to spot outages, slow CI, flaky tests and Dev box failures. Front-end time spent waiting on machines P90 for webpack falls from ~90 minutes to ~3–5 after swapping bundlers. A paper cuts button captures ~1,500 reports from ~500 engineers; ~100 low‑effort fixes ship. Bazel remote execution, Dev boxes, GitHub Enterprise and shadowing use drive prioritization toward CI, review turnaround and focus‑time improvements.
AI-written summary · about
chapters (18)
- 0:00Developer productivity intro
- 2:09Developer productivity lowercase
- 3:12Developer productivity tools
- 7:05Tracking bad days
- 10:53Slow tooling measurement
- 13:42Time waiting on machines
- 14:44Metrics across tools
- 16:50Introducing Paper Cuts
- 18:56Paper Cuts lessons
- 21:43Next-gen Paper Cuts
- 22:26Shadowing engineers
- 23:50SPACE framework adoption
- 26:21Focus time & collaboration
- 28:50Simplified SPACE metrics
- 30:15Improving sentiment & focus
- 31:41Focus time tools
- 33:45Key takeaways
- 35:50Q&A
related talks
27mDeveloper Experience: It’s Time to Start Complaining
44mSPACE: Measuring Developer Productivity
28mHow we do developer productivity at Slack
58mDeveloper Joy – How Great Teams Get S%*t Done
36mDevOps Performance From a Different Dataset: What 30M Workflows Reveal
06mDevops Behavioral Economics: Displayed...
43mMeasuring Developer Productivity: The Good, the Bad and the Data
30mOptimizing for Time: Dark Matter