Published: 12 August 2026
Food safety culture used to be the part of an audit where everyone talked in generalities. It is now an explicit requirement in GFSI-benchmarked standards and a subject of Food Standards Agency guidance, which means auditors arrive looking for evidence rather than intentions.
The problem for most technical managers is that culture sounds like something you demonstrate through posters, briefings and an annual survey. In practice, the most revealing evidence sits somewhere much more concrete: what happens in your plant on the day a procedure changes.
"Tell me about your food safety culture" is hard to answer well. "This allergen SOP was revised in April — show me the training records for everyone who runs that line" is easy to answer badly. The second question is the one that actually tests the first.
Why culture became something you have to evidence
The shift makes sense once you look at what causes food safety incidents. Very few are caused by a business not having a procedure. A great many involve a procedure that existed, had been revised, and was not reliably followed by the people doing the work — because they had been told about it once, at handover, in a language that was not their first, on a shift where two people were off.
Auditors and regulators responded by looking past the documentation at whether the business behaves as though the documentation matters. Communication and training are where that becomes observable, and change is where it becomes measurable.
The change moment is your best and worst evidence
Consider the ordinary sequence in most plants. A customer specification changes, or an allergen is introduced on a line, or a sanitation chemical is swapped. The technical team updates the SOP and issues it through document control. A supervisor mentions it at handover. Someone puts a printed copy in the line folder.
Six months later an auditor asks which version each operator was trained against. The document system knows the revision history. The training system, if there is one, has a generic allergen awareness module completed eighteen months ago. Nothing connects them. The plant has probably been running correctly the whole time, and there is no way to show it.
That gap is where non-conformances are raised against businesses that were doing the job properly. It is also, read the other way, the strongest culture evidence available — because a documented chain from revision to retrained, verified operator is difficult to fake and easy to inspect.
The hard part is working out who is affected
The reason most businesses fall back on a verbal briefing is that identifying the affected population is genuinely difficult without a mapping.
An allergen SOP revision does not affect everyone on site, and it does not affect a neat department either. It affects the operators on the lines where the allergen is present, the sanitation team responsible for changeover, the goods-in team handling the ingredient, the QA technicians verifying the checks, and the shift supervisors overseeing all of it — across however many shifts you run. Some of those people are agency. Some transferred from another line last month.
Sending it to everyone looks safe and is not: completion rates fall, the people who genuinely needed it are lost in the noise, and the record shows a site-wide assignment rather than targeted retraining. Sending it to a department misses the people who move between them.
What the mapping needs to hold
| Dimension | Why it matters |
|---|---|
| Document to process | Which SOPs, work instructions and prerequisite programmes govern each process step |
| Process to line | Allergen and equipment exposure differ by line, so the affected population does too |
| Line to shift | The same line has different people on it depending on the shift pattern |
| Role to task | An operator, a sanitation technician and a QA checker need different things from the same revision |
| Employment type | Agency and seasonal workers carry the same obligation and are the most commonly missed |
Build that mapping once and a revision becomes an assignment with a due date. Skip it and every revision is a judgement call made under time pressure by whoever is on shift.
The language problem is a food safety problem
Large parts of the UK food manufacturing workforce do not read English as a first language. This is treated as an HR consideration and it is actually a control failure risk: if the person performing a critical control point cannot fully follow the instruction they were trained on, the training did not do its job regardless of what the completion record says.
Two things follow. Training on procedures that carry food safety consequences should be available in the languages your crews actually read, and completion should be tracked per person across languages so reporting does not fragment. An auditor asking about a named operator should not need to know which version of the module they took.
If your last three SOP revisions were communicated only in English, and a meaningful proportion of the affected operators read another language first, your training records and your actual position on the floor may not match.
A knowledge check is not verification
There is a useful distinction between two things that both get called training records.
- A knowledge check shows the operator understood the revised procedure. It is quick, it scales, and it can be delivered at handover on a kiosk or a phone.
- Supervisor verification shows the operator performs the task correctly under the revised procedure. It requires someone watching, and it is the evidence that carries weight when the task is a control point.
Most sites do both and record only the first. The observation happens — a supervisor watches the changeover, satisfies themselves, and moves on — but it leaves no trace. Recording it against the operator, with the observer, the date and the document version, converts routine supervision into audit evidence at almost no additional cost.
Five questions worth rehearsing
If you can answer these from a system rather than from memory, your culture evidence is in reasonable shape.
- Which version of this SOP was this named operator trained against, and when? The version, not just the topic.
- When you revised it, who was reassigned, and who is still outstanding? The outstanding list is the interesting half.
- Who is currently cleared to perform this task on this line? By task, not by job title.
- Show me the same records for an agency worker who was here in March. Retention after they leave is routinely tested.
- Who verified that this operator could do it, and on what date? Verification separate from completion.
None of these require a new quality system. They require the link between document control and training records to exist as data rather than as an assumption held by the technical manager.
Where to start
Take the most recent SOP revision that carried real food safety consequences. Work out, by hand, exactly who should have been retrained. Then compare that against who actually was. The size of the difference tells you whether this is a documentation exercise or a genuine gap — and either way, that mapping is the foundation you build the rest on.
Frequently asked questions
What is food safety culture in an audit context?
It is the shared attitudes, values and behaviours around food safety within a business. It has moved from an informal concept to an explicit element of GFSI-benchmarked standards and FSA guidance, which means auditors look for evidence of it rather than assertions about it. Communication, training and how the business responds to change are among the most visible indicators.
Does a Level 2 food hygiene certificate cover our training obligation?
Not on its own. Food law requires food handlers to be supervised and instructed or trained in food hygiene matters appropriate to their work activity. A general certificate does not evidence that a named operator was trained on your specific procedure for the specific task they perform.
How should we evidence retraining after an SOP change?
With a record showing which version of the document each person was trained against, the date, and who verified their competence. The strongest version of this also shows the reassignment list the change generated and who remained outstanding, because that demonstrates the process worked rather than just that some training happened.
Do agency workers need the same records?
Yes. Induction and task-specific training records for agency and short-tenure workers are routinely examined, and the fact that someone worked for four weeks does not remove the obligation. Retaining those records after they leave matters, because an audit may cover a period during which they were on site.
Sources & further reading
- Food Standards Agency — Food safety culture guidance for food businesses
- Food Standards Agency — Allergen guidance for food businesses
- BRCGS — BRCGS Global Standard for Food Safety