Evaluating Engineers — Calibrating Impact, Not Output
Evaluate engineers on scope of impact, autonomy, and growth — not lines of code, PRs merged, or hours worked.
When to use
- Performance reviews and promotion calibration
- Hiring debriefs
- Identifying high-potential engineers early
Tradeoffs
- Leveling frameworks create gaming behavior (engineers optimize for the rubric)
- Under-leveling demotivates; over-leveling sets engineers up to fail
Leveling Matrix
| Dimension | L4 (Senior) | L5 (Staff) | L6 (Principal) |
|---|---|---|---|
| Scope | Team | Cross-team | Org-wide |
| Decision-making | Executes decisions | Drives decisions | Shapes strategy |
| Ambiguity | Structured problems | Ambiguous problems | Creates clarity for others |
| Mentorship | Helps teammates | Grows the team | Raises the bar across org |
| Delivery | Delivers features | Delivers projects | Delivers programs |
Calibration Bias Checklist
- Recency bias: don't let the last month overshadow the full review period
- Halo effect: strong performance in one area inflating all dimensions
- In-group bias: engineers who communicate in your style score higher unconsciously
- Availability bias: who you can remember most easily ≠ who performed best
Gotcha: Promotion decisions happen in calibration sessions with peers, not in your 1:1. Coach your report on what the committee will see — their peers' perception of their impact, not just what they're delivering.