Executive summary
S$8,000
fine issued to a Singapore caterer after 395 people fell ill, a case a temperature sensor would not have caught
Key conclusions
- Codex Alimentarius treats monitoring and record-keeping as two of HACCP's seven founding principles. Poor records can invalidate an otherwise sound plan.
- An automated CCP feed fixes missing or incomplete monitoring records. It does not address food safety culture, allergen management or food fraud.
- Automate the record and leave judgement to people. A data logger cannot replace the walk-through inspection that a supervisor or auditor carries out.
At most food processing plants, staff still handwrite critical control point records on a laminated sheet. They do so even where the equipment already has a digital sensor producing the same reading. This gap between what is measured and what is recorded matters. Codex Alimentarius treats monitoring and record-keeping as two of HACCP's seven founding principles. Its guidance is direct about the consequence of poor record-keeping: poor records can nullify a scientifically sound, fully validated plan. Without good records, nothing shows an auditor, a regulator or a court that the monitoring took place.
In the plants I audit, the same pattern repeats. A chiller with a digital display sits beside a clipboard. Whoever is on shift fills in the clipboard, at whatever interval the shift allows. Nobody reconciles the two numbers, because nothing requires it until an auditor asks.
Automated capture fixes some recurring audit findings but not others
SGS's 2026 review of FSSC 22000 audits and FSNS's ranking of recurring non-conformities point to the same small set of categories. Automated CCP monitoring closes some of them and does not address the rest:
| Recurring finding | Does an automated CCP feed fix it? |
|---|---|
| Missing or incomplete records for devices monitoring or controlling a food safety hazard (FSNS's second most-cited non-conformity) | Yes. This is precisely what a logged, timestamped feed replaces. |
| Overdue calibration on monitoring instruments with no documented risk assessment | Partly. The logger can flag its own due date; it cannot decide whether the risk assessment was adequate. |
| Documentation not updated to the current scheme version | No. That is a document-control gap. |
| Food safety culture | No. Culture is a management and training question a sensor has no view on. |
| Allergen management, food fraud and food defence | No. These turn on procedure, labelling and intent. |
A logged CCP record brings four specific benefits
- The number an auditor reviews is the number the instrument produced. Nobody transcribes it from memory at the end of a shift.
- Every reading carries a timestamp. This prevents the common failure in which one entry covers eight unmonitored hours.
- The system flags a missed reading as a gap, so an auditor does not have to spot a blank cell.
- The monitoring device itself becomes part of the same maintenance record. This replaces the separate spreadsheet that nobody reconciles, and addresses FSNS's own second-ranked finding.
Sensors would not have caught the lapses in a recent Singapore case
In April 2025, Singapore's Food Agency fined Shiok Kitchen Catering Pte Ltd S$8,000. The fine followed a joint inspection with the Ministry of Health, after 395 people fell ill with gastroenteritis from food the company catered in November and December 2023. Inspectors found a dirty chiller, a dirty ice-making machine, cockroaches in an unused chiller and a sample of ready-to-eat spinach unfit for consumption. A temperature sensor was not needed to find any of these problems. Finding them required someone to inspect the premises, recently enough for the record to mean something.
This case shows the limit of automated records. A logger records what happened. It does not decide whether a reading should have triggered corrective action, and it cannot replace the walk-through that a supervisor or auditor carries out. Automate the record, leave judgement to people, and expect automation to reduce one category of audit finding.
