Ask ten food manufacturers to show you their validation records and you will get metal detector check sheets, calibration certificates, and CCP monitoring logs. Those are good records. None of them is validation.
The confusion between validation and verification is one of the most common technical misunderstandings in HACCP systems, and it is not a semantic problem. It creates a specific, dangerous gap: a control measure that is diligently monitored and thoroughly verified, but was never proven capable of controlling the hazard in the first place. The system generates a large volume of evidence that the control is being operated, and no evidence at all that operating it achieves anything.
The distinction in plain terms
Three activities are involved, and they answer three different questions.
Validation asks: will this control measure work? It is the process of obtaining evidence that a control measure is capable of controlling the hazard to the specified outcome. Validation happens before the control is relied upon, and again whenever something material changes.
Monitoring asks: is it working right now? It is the planned sequence of observations or measurements at a CCP to determine whether the process is operating within critical limits, in real time, so that loss of control is detected as it happens.
Verification asks: has it been working? It is the application of methods, procedures and tests — in addition to monitoring — to confirm that the system is operating as intended. Verification is retrospective and periodic.
A simple way to hold the difference: validation proves the critical limit is the right limit. Monitoring confirms you are within it. Verification confirms that the monitoring itself is being done properly and that the outcome is what you expected.
Where validation actually comes from
Validation evidence does not come from your own monitoring records. This is the point most systems miss. Your temperature logs show that you held 72°C for 15 seconds. They do not establish that 72°C for 15 seconds delivers the required log reduction for the pathogen of concern in your specific product matrix.
Acceptable sources of validation evidence include:
- Scientific and technical literature — published thermal death time data, validated process parameters for comparable products
- Regulatory and industry guidance — established guidance documents specifying validated process criteria
- Challenge testing — deliberate inoculation studies, conducted by a competent laboratory, on your product
- Predictive microbiological modelling — validated models applied within their stated limits of use
- Historical in-plant data — where sufficient, statistically meaningful data exists from your own process
- Expert advice — documented opinion from a suitably qualified authority, with the basis stated
Whichever source is used, the validation record must connect the evidence to your product, your process, and your equipment. A thermal process validated for a thin liquid does not validate the same time and temperature applied to a viscous, particulate-containing product where heat penetration behaves differently.
What verification looks like when it is done properly
Verification is broader than most sites treat it. A complete verification program typically includes:
- Calibration of monitoring equipment, at a defined frequency, traceable to a reference standard
- Record review — a second person confirming that monitoring records are complete, within limits, and signed at the time of the activity
- CCP record review specifically, conducted before product release where the CCP governs safety
- Product and environmental testing, used as a verification tool rather than as a release test
- Internal audit of the HACCP system against the plan
- Review of deviations, complaints and recalls for signals that a control is drifting
The critical discipline is that verification must be performed by someone other than the person who did the monitoring. Self-verification is not verification; it is monitoring performed twice.
When revalidation is triggered
Validation is not a one-time exercise filed at implementation. The plan must be revalidated when:
- The product formulation changes — particularly pH, water activity, salt, or preservative system
- A raw material or supplier changes in a way that affects the hazard profile
- Process equipment is replaced, modified, or its capability changes
- Throughput or batch size changes in a way that affects process delivery
- New scientific information emerges about the hazard or the control
- Deviation trends suggest the control is operating at the edge of capability
A validation record dated at system implementation, with three equipment changes and a reformulation since, is not a current validation. It is a historical document.
The audit consequence
A HACCP plan is worth testing with two questions about every CCP. First: what evidence shows that this critical limit controls this hazard in this product? Second: what evidence shows that it has been doing so? Most systems answer the second question confidently and the first with monitoring records — which is the gap.
If you can only answer one of those questions, the one you cannot answer is the one that matters most, because it determines whether all that monitoring evidence means anything at all.
Practical takeaways
- Separate your files. Keep validation records physically or digitally separate from monitoring and verification records. If they live in the same folder, they will be treated as the same thing.
- Write the validation basis on the HACCP plan itself. Next to each critical limit, state the source of the validation evidence and its date.
- Test it with one question. For each CCP, ask: if a customer asked why this limit and not a lower one, could I answer with evidence that is not my own monitoring data?
- Define your revalidation triggers in writing. List the specific changes that require revalidation, so it is a procedural trigger rather than a judgement call.
- Check the independence of verification. If the operator who monitors also signs the verification review, the verification step is not doing its job.
Validation proves the control is right. Verification proves you did it. You need both, and the absence of the first quietly undermines everything built on top of it.