CPGP common mistakes
CPGP · Understand the traps and how to avoid them.
Using the final test to excuse an uncontrolled process
A sample result does not establish that every process condition or product requirement was met.
What to do instead
Review the complete record and investigate departures before the authorized disposition.
Changing a limit after seeing a failed result
A retrospective limit can conceal the actual deviation and invalidate the intended comparison.
What to do instead
Use approved criteria and document a justified change through the appropriate process.
Assuming three successful runs prove every future condition
A limited demonstration does not cover unstudied loads, changes or operating ranges.
What to do instead
State the study design, covered conditions and continuing monitoring responsibilities.
Calling a completed correction proof of effective corrective action
Fixing the immediate item may leave the cause and recurrence pathway untouched.
What to do instead
Identify the supported cause and evaluate the implemented action against a planned effectiveness check.
Treating a supplier certificate as unlimited permission to use materials
Supplier information does not remove every receiving, identity, qualification or status-control responsibility.
What to do instead
Apply the actual approved material controls and relevant regulatory requirements.
Averaging away an excursion
An acceptable average can hide an individual value outside a specified condition.
What to do instead
Assess the individual results, acceptance rule and affected time or product scope.
Rewriting the original record during an investigation
Replacing the original observation destroys the evidence needed to understand what happened.
What to do instead
Preserve the original and make attributable corrections with a reason and audit trail.
Transferring another product cycle without an assessment
Materials, packaging, microbial challenge and load configuration can change the process outcome.
What to do instead
Evaluate compatibility and validation coverage before applying an existing process to a changed product.