The summary.
SEVERITY IS TECHNICAL IMPACT; PRIORITY IS URGENCY
They differ, and conflating them causes bad decisions. Always treat data loss, security exposure and money handled wrongly as critical — fixing the code does not undo them.
REPORT ONE PROBLEM PER ITEM, WITH EXACT STEPS AND EXACT DATA
Many defects depend on specific values, and multiple problems in one item cannot be triaged separately.
THE ESCAPE IS MORE INFORMATIVE THAN THE DEFECT
Ask why it reached this stage undetected, add a test that would have caught it, and categorise causes over time to find where the process leaks.
Blame produces concealment, which prevents learning entirely.
QUESTIONS ASKED BEFORE BUILDING PREVENT DEFECTS ENTIRELY
Which is the highest-value thing a tester does, and why testers belong in planning rather than at the end.
NEVER DISABLE A FAILING TEST TO UNBLOCK THE PIPELINE
The next real failure becomes invisible.
MEASURE DEFECTS THAT ESCAPED, NOT DEFECTS FOUND
Counting defects found encourages reporting trivia and adversarial relationships. Never report a percentage implying completeness.
