How do you measure the effectiveness of your testing efforts?

Test effectiveness is hard to measure because the thing you want — "we caught the bugs that matter" — is partly counterfactual. The honest answer is to triangulate.

Outcome metrics — what stakeholders care about:

• Defect escape rate: defects found by customers / total defects. Lower is better. The single best signal of test quality.
• Defect injection rate: defects per release / size of release. Trends over time matter more than absolute numbers.
• MTTR for production bugs: how quickly do we detect, diagnose, fix? Long MTTR points at instrumentation and observability gaps.
• Customer-reported severity profile: are bugs reaching production trivial or critical? Critical-severity escapes are the actionable signal.

Process metrics — leading indicators:

• Coverage (line, branch, behaviour) — useful as a floor, not a target. 80% line coverage of trivial code is meaningless.
• Test cycle time — how long from "build ready" to "shipped"? Long cycle = bottleneck.
• Automation ratio: % of regression run automatically.
• Flake rate — high flake means low trust means escape rates climb.

What I'd avoid relying on: number of tests written (gameable, says nothing about quality), pass rate (should always be 100% on main; anything else is a trust signal not a quality one), bug counts as a positive ("we found 200 bugs!" might mean QA is heroic, or that the build was broken — without context it's noise).

The senior signal: presenting metrics as a triangulation, knowing the limits of each, and tying them to outcomes the business actually cares about.