In 30 seconds
- Counts defects that slipped past the team's checks.
- A direct signal of quality as users experience it.
- Each one is a lesson for testing and the Definition of Done.
Tracking them
- Count defects reported after release, per release or month
- Record severity and where they were found
- Note which check should have caught each one
What they reveal
Escaped defects show gaps in testing, unclear acceptance criteria or a weak Definition of Done. Patterns matter more than counts: the same kind of bug escaping repeatedly points to a fixable gap.
Reducing them
- Add a test for every escaped bug before fixing it
- Strengthen automated testing where escapes cluster
- Discuss notable escapes in the retrospective
Myth vs reality
Myth · Zero escaped defects is realistic.
Reality · Some will escape; aim for fewer, less severe ones.
Myth · Escapes are the testers' fault.
Reality · Quality is the whole team's responsibility.
Myth · More releases means more escapes.
Reality · Smaller releases usually mean fewer.
The AI angle
Watch escaped defects closely after adopting AI tools. A rise often signals that review and testing haven't kept pace with the volume of generated code.
FAQ
How is this different from defect density?
Escaped defects count what users found; density relates defects to size.
Should low-severity escapes count?
Track them, but focus on severity.
What's a good target?
Aim for a falling trend, especially in severe escapes.