Proving your cold chain: what to measure, and how

"It looked like it had been respected" no longer satisfies an inspector, nor a professional customer who demands proof. A cold chain is proven with dated, continuous, exportable records — not with a thermometer read once a day and written down from memory. This guide covers what to measure, where, and how to keep the proof itself from being contestable.

Thresholds by product family

Regulatory thresholds vary significantly with the nature of the product — copying one family's threshold onto another is the most frequent mistake among operators broadening their range:

• Fresh dairy: 4 °C maximum in general, with the exact threshold set by order according to product type.
• Fresh meat: 4 °C, with stricter limits for offal (3 °C) and poultry (4 °C as well, but a finer inspection tolerance).
• Fresh fishery products: close to 0 °C, on melting ice — the tolerated margin of error is the narrowest of all families.
• Frozen goods: −18 °C, with a one-off tolerance of +3 °C only during short handling operations (loading, shelf display).
• Ready-to-eat cut fruit and vegetables: 4 °C, like fresh dairy.

These thresholds are regulatory maximums, not targets to approach as closely as possible — a product held permanently at 3.9 °C when the limit is 4 °C leaves no margin for a micro-break during handling.

Where to place sensors: the frequent blind spots

A properly regulated cold room does not guarantee an intact cold chain — the break most often happens elsewhere:

• The loading dock: the moment a product moves from the cold room to the vehicle is the most frequent break window, and the least often measured. A sensor at the dock, not only in the cold room, reveals what the rest of the chain does not show.
• The delivery vehicle: the temperature of the refrigerated compartment does not always reflect that of the product itself, especially on a round with several stops and door openings. A sensor in contact with the product, not only in the vehicle's ambient air, gives a truer measurement.
• The chilled display in store: a warm spot can exist at the edge of the unit or near a frequently opened door, while the central sensor shows a correct temperature. Several measurement points beat a single one at the centre.
• Temporary storage between two steps — a pallet left on a dock awaiting loading, even for ten minutes in high summer, can be enough to push a sensitive product out of its range.

Alerting before the break rather than recording it afterwards

A record that merely logs temperature only documents a break after it has happened. An alert threshold set ahead of the regulatory one — a notification at 6 °C for a regulatory limit of 4 °C, say — leaves time to intervene before the product actually leaves its storage range. That is the difference between a system that prevents a non-conformity and one that merely observes it.

What an inspector looks at first

1. The continuity of the history — a record with gaps (sensor unplugged, export missing for several days) draws more attention than a documented isolated deviation.
2. Sensor calibration — a last-calibration date available, not merely a reading displayed without reference.
3. The handling of deviations — a threshold breach with no associated corrective action weighs more heavily than a breach followed by a traced decision (product set aside, returned, destroyed).

Common mistakes

• A sensor never calibrated since installation, or whose calibration is undocumented. A sensor with a constant 1 or 2 °C offset silently invalidates months of records.
• A history that cannot be exported in the format an inspection asks for — a system that draws a curve on screen but produces no dated, timestamped export is not usable proof.
• A single measurement point for a heterogeneous space — a truck, a display unit or a large cold room can hold several different temperature zones.
• Widely spaced manual readings — a reading every four hours does not catch a thirty-minute break during a failure or a door left open.

Checklist

1. Identify the exact regulatory threshold per product family, not a generic one.
2. Place a sensor at the loading dock, not only in the cold room.
3. Measure in contact with the product in the vehicle, not only the ambient air.
4. Configure an alert ahead of the regulatory threshold, not only a passive recording.
5. Document sensor calibration and its date.
6. Check that the history exports, dated and timestamped, in a few minutes.