Agriculture & Future Food Systems · Article

Verification, audits and records

A reviewer of a grain-processing line needs more than completed forms: observations, instrument checks and linked records must answer whether the controls operated as intended.

Preparation and related reading

Shared Foundations

This article helps you

Distinguish routine monitoring from verification of its implementation and suitability.
Use a record to reconstruct an actual control event.
Interpret completeness counts without turning them into a food-safety percentage.

Look beyond the completed form

Imagine a grain-processing line with a completed monitoring form for every shift. A verifier watches one check and discovers that the recorded observation comes from a different point from the one specified in the plan. The paperwork can be complete while its meaning is wrong.

Monitoring follows the relevant control during operation. Verification uses additional checks to examine whether the system follows its plan and controls hazards as intended. These activities should support timely action rather than merely produce documents. The grain-line example has no approved treatment or inspection frequency; it illustrates how to question a record.

Ask a different question

An operator may record the specified condition while running the process. A verifier asks whether the correct method was used, whether the instrument is suitable and whether the plan remains adequate. Observing the task, checking instruments, reviewing records and appropriate targeted testing can answer different parts of that question.

No single activity covers every part of verification. A calibrated instrument at the wrong location may not represent the relevant condition. An appropriate location with unreliable measurement also leaves a gap. Match the verification activity to the question rather than assuming the word audit guarantees that the technical issue was examined.

Use competent, sufficiently independent checks

People conducting verification need appropriate technical competence. For checks of monitoring and corrective action, use someone other than the person responsible for performing those tasks; relevant external expertise may be needed where capability is unavailable within the business. Independence does not remove the need to understand the process.

In the imagined line, a separate competent reviewer observes the actual measurement and examines its connection to the plan. A purely administrative signature cannot establish that connection. An audit can identify findings and the evidence behind them; its value depends on scope, method and follow-up, not on the prestige of a logo.

Reconstruct a control event

A useful record identifies the product or lot, time, actual observation, method or instrument where relevant, responsible person and response. Link a deviation to the affected product account and its disposition. These connections let another person reconstruct what happened instead of guessing from a tick.

Compare an invented note, line checked, with a record that identifies the specified check, its result and the action taken. The fuller record can support review even if it is a simple paper form. An electronic system is not automatically more accurate: missing identity, incorrect time or copied values can still distort the event.

Interpret completeness counts honestly

Suppose a hypothetical review finds that 96 of 100 expected records contain the required identifiers, observation and signature. Its completeness result is 96%, with four records lacking information. This is a document-quality observation, not a measurement that 96% of food is safe.

The diagram separates those records from the operational conclusion still needing evidence. The missing four might concern different conditions or periods, so assess their relevance rather than assigning them the average status of the others. Likewise, a complete form can contain an incorrect observation. Counting fields is useful, but it cannot settle accuracy or technical adequacy.

Complete forms are one kind of evidence

Original hypothetical completeness count, not an audit pass standard or proof of product safety. No quantitative size scale is implied.

Read the diagram as text

In an invented 100-record review, 96 contain the specified fields and four do not. Both the incomplete records and the accuracy of completed observations need appropriate evaluation.

Parts

100 expected records: Stipulated document review
96 contain required fields: 96% completeness, not a safety percentage
4 lack information: Relevant periods and implications need assessment
Check observation accuracy: Method, location, instrument and actual activity
Investigate the gap: Do not infer a result for an unrecorded event

Connections

100 expected records → 96 contain required fields: field review
100 expected records → 4 lack information: field review
96 contain required fields → Check observation accuracy: additional check
4 lack information → Investigate the gap: bounded finding

Close the finding and review change

For the wrong measurement point, establish the relevant affected period, correct the practice and evaluate the impact on the control evidence. Verify that the revised check is actually performed as intended. A finding marked closed should have a basis, not simply a promised training session.

Review the plan when relevant changes or repeated failures raise questions about adequacy. Preserve monitoring, correction and verification records so the history remains understandable. Verification strengthens a working system by testing its assumptions and implementation; it does not retrospectively create evidence that was never recorded or authorize product release by document count.

Worked examples

Compute only the record result

Ninety-six of 100 stipulated expected records have the required fields. What does the result describe?

Completeness = 96/100×100 = 96%.
Missing or incomplete records = 100−96 = four.
Accuracy and the food-control outcome require additional evidence.

96% document completeness in this hypothetical review, not 96% safe product.

Examine a complete but unsuitable observation

Every grain-line form is signed, but the observer uses a different measurement point from the plan. Does signature completeness close the finding?

A signature identifies a recorded action.
The action’s location differs from the specified method.
Assess the method, relevant product history and necessary follow-up; then check correct implementation.

No. The forms may be complete while the control evidence is unsuitable.

Check your understanding

Does an electronic form automatically contain an accurate observation?
Show answer and reasoning
No.
Format does not establish the value, location, time or identity.
Does being a different reviewer automatically provide the needed technical competence?
Show answer and reasoning
No.
Independence and relevant competence are separate requirements.
Does a promise to retrain establish that a recurring finding has been effectively resolved?
Show answer and reasoning
No.
The changed practice and its relevant outcome still need checking.

Key ideas

Verification asks about implementation and adequacy beyond the routine reading.
A record must retain the meaning of the observation.
Completeness, accuracy and food safety are different conclusions.

References

Sources supporting the explanations and examples in this article.