Accelerated or real-time? Choosing the right shelf-life study
Accelerated testing gets you to market faster. It also lies about certain failure modes. Here is how to decide which study your product actually needs.
Read the article →Almost every plant we audit has a HACCP file. Rather fewer have a HACCP system. The difference shows up within twenty minutes of walking the floor, and it comes down to three things.
Failure one: the plan does not describe the plant
Templates downloaded or bought from a consultant describe a generic process. The auditor walks the line, finds a step that is not on the flow diagram, and the credibility of the whole document is gone. The flow diagram must be verified on the floor, by walking it, and updated whenever the process changes.
Failure two: too many critical control points
A plan with fourteen CCPs is a plan where nobody monitors any of them properly. Genuine CCPs are the few steps where loss of control leads directly to an unacceptable safety risk and no later step will catch it. Everything else is a prerequisite programme or an operational control. Fewer, well-monitored CCPs beat a long list.
Failure three: records without decisions
Monitoring records signed at the end of the shift, all values suspiciously identical, with no evidence of what happened when a reading was out of limits. Auditors specifically look for a deviation and its corrective action. A file with no deviations in twelve months is not evidence of control; it is evidence that nobody is recording honestly.
What a working system looks like
- A verified flow diagram matching what actually happens on the floor.
- A hazard analysis with reasoning recorded, not just conclusions.
- Two to five real CCPs with measurable critical limits.
- Monitoring at a stated frequency, by a named role, recorded at the time.
- Documented corrective actions, including product disposition, for every deviation.
- Verification activities and an annual review with evidence.
Train the people, not the file
Auditors ask operators, not managers, what they do when a reading goes out of limits. If the operator can answer clearly, the system is real. If they look towards the quality manager, it is not. Training and a short, readable work instruction at the control point matter more than the thickness of the manual.