hello@sterling-grc.com +1 201 263 4443 New York ISO & GRC consulting in 60+ countries
SterlingInternational Consulting Talk to a consultantTalk to us
Clause 8.9.3ISO 22000:2018 · Control, evaluate and decide safely

ISO 22000 clause 8.9.3: prevent food safety control failures from recurring

ISO 22000 clause 8.9.3 requires evaluation of the need for corrective action when CCP critical limits or OPRP action criteria are not met. The organisation must maintain arrangements to identify and eliminate causes, prevent recurrence and return the process to control. All corrective actions need retained documented information, including results and verification of effectiveness.

What clause 8.9.3 requires in practice

Clause 8.9.3 asks the organisation to judge, each time a CCP limit or OPRP criterion is missed, whether corrective action is needed. It then wants documented information that defines the actions used to find and remove the cause, stop the problem coming back and put the process back under control once a nonconformity has been recognised.

Those actions are expected to include looking at nonconformities raised through customer or consumer complaints and regulatory inspection reports, and at trends in monitoring results that could signal a loss of control. They also include establishing causes, deciding and carrying out measures that prevent recurrence, writing down what was done, and checking that it worked. All corrective actions leave retained documented information. In short, the clause separates the quick fix from the lasting one: restoring the setting is correction under 8.9.2, whereas understanding why the setting drifted and proving the remedy holds belongs here.

Use more than the latest failed result

Review relevant nonconformities from customer or consumer complaints and regulatory inspection reports. Examine trends in monitoring results that may indicate loss of control. A pattern can reveal a weakness before or beyond a single isolated failure.

Define the problem clearly: what requirement was not met, where, when and with what consequence. Separate confirmed facts from explanations that still need investigation. This helps avoid selecting a familiar remedy before understanding the cause.

Immediate correction and product control should proceed as required. Investigation must not leave affected food uncontrolled while the team debates a longer-term solution.

ISO 22000 consulting & certification support

Get clause 8.9.3 right the first time — with a global leader in ISO 22000 implementation

A Sterling consultant works the requirement through with your team, links it to the rest of the system and carries it to certification readiness. Expert and experienced consultants, 8,100+ client engagements in 60+ countries.

Why preventing recurrence protects the business

A food business that fixes the same failure every month is paying twice: once for the lost or reworked product and again for the risk that one repeat reaches a customer. Customers and assessors read repeat failures as a sign that management does not understand its own process. Corrective action that finds a real cause, such as an unclear instruction or a gap between two teams, removes that recurring cost and builds credibility. Using complaints, inspection findings and monitoring trends as inputs also means weak signals are caught before a serious event. When effectiveness is checked accurately, management can see which actions deserve confidence and which were only reported as complete.

Determine causes that explain the evidence

What counts as evidence

Examine the process, equipment, instructions, competence, resources and interfaces relevant to the event. Human action may be part of the sequence, but “operator error” often fails to explain why the arrangement allowed the error or failed to detect it.

Use suitable investigation methods. A cause-and-effect discussion, process review or simple comparison of conditions may help. No single branded tool is required by ISO 22000; the method needs to produce an evidence-based explanation.

Check whether the proposed cause fits the facts and whether more than one cause contributed. A conclusion selected only because it is easy to fix can leave the actual weakness untouched.

Implement action and verify effectiveness

Choose action that addresses the identified cause and prevents recurrence. Assign responsibility and determine how completion and effectiveness will be evaluated. Restoring a setting or replacing a component may be necessary but may not address why the failure arose.

Document the results of action and verify that it is effective. The check should be capable of detecting recurrence or continued weakness under relevant conditions. A signed completion box proves an action was reported complete, not that it worked.

Use the results in performance evaluation and updating where relevant. Repeated failures after closed actions deserve reconsideration of the investigation and remedy.

Illustrative food-manufacturing example

Illustrative example 1

A manufacturer experiences repeated failures associated with a process changeover. Initial responses restore the process each time. A broader investigation finds that the changeover instruction omits a relevant setup condition and that verification does not check it.

Illustrative example 2

The team revises the arrangement and evaluates performance across subsequent relevant changeovers. Training supports the change, but the action is not described merely as reminding operators to be more careful.

Automate™ on Effivity · included in every Sterling engagement

Keep clause 8.9.3 alive in Effivity

Record what you decided under clause 8.9.3, link it to the records and controls it affects, and let the review date trigger the next look. Effivity is the ISO & GRC software used in 50+ countries — and part of every Sterling ISO 22000 programme.

Illustrative catering example

A caterer receives recurring complaints that raise a food safety concern around a service handover. The investigation compares the documented arrangement with actual work and identifies a responsibility gap between two teams.

The organisation clarifies the handover, develops the affected personnel and checks later performance using relevant evidence. It also evaluates affected product and communication needs separately from the long-term corrective action.

Evidence and common weaknesses

Common weaknesses

Retain the problem description, relevant investigation, causes, actions, results and effectiveness verification. The record should let another competent person understand why the action was selected and what establishes that it worked.

Common weaknesses include automatic retraining, causes unsupported by evidence, action closure before implementation and effectiveness checks unrelated to the original problem. Another is overlooking adverse trends because each individual result remains within a reporting threshold.

This operational corrective-action requirement connects with the broader FSMS nonconformity process; a coherent system can serve both while preserving their specific needs.

Frequently asked questions

Does every failure require the same investigation depth?

No. Use a suitable approach for the event and its implications, while meeting the requirement to evaluate corrective-action need and address causes appropriately.

Is retraining a corrective action?

It can be where a competence gap is supported by evidence. It is weak when used without understanding the cause.

When is an action effective?

When suitable evidence shows that it addresses the intended cause and control need. The check must relate to the original problem.

Strengthen an investigation that keeps returning

Discuss a recurring food safety nonconformity with Sterling. A scoped review can connect causes, action design and suitable evidence workflows through Effivity ISO & GRC automation software. Explore Root Cause Analysis & CAPA, the nonconformity guide and the hub.

Source basis and reading note. This is original practical guidance for ISO 22000. Use an authorised copy for definitive requirements. Illustrative examples describe hypothetical situations, not client results. Implementation methods are distinguished from requirements; independent bodies decide management-system certification outcomes. Issuer standard record: iso.org.
Disclaimer. This guide is provided for general information only. It is not the text of the standard, not legal or professional advice, and not a representation of how any certification body will assess your organisation. Sterling International Consulting accepts no liability for any loss arising from reliance on it. Requirements are applied to each organisation's own context through an engagement; contact us to implement ISO 22000 in your organisation.