Long-form writing from the team and our customers. What we measure, why we explicitly don't measure other things, and how the best engineering managers actually run their week.
Every workforce-analytics tool ships a screen that looks like a leaderboard. We removed ours in v0.1.3 and we've never been asked to put it back. This essay is the argument we made internally — and the data that ended the debate.
Our data manifest in plain English. Six things every other tool in our category collects, and the exact internal arguments that ended each one.
Drawing 200 sparklines at 60fps on a 4-year-old laptop sounds easy. Turns out the bottleneck wasn't the render — it was the data transform. A teardown.
Anonymised aggregate data from 412 customer fleets reveals a remarkably consistent shape to the workday. Plus three patterns that genuinely surprised us.
How one 410-person bio-tech replaced their weekly all-hands status meeting with a 2-minute scan of Overview — and reclaimed 23 person-hours every Monday.
The single most common pattern across our customer base is also the most fixable. A short essay on why "they worked late again" should always be a manager problem first.
How we built an agent that runs at <2% CPU, has a ~28MB footprint, and only writes data the person being measured can also see. Diagrams included.