A pallet of vaccines leaves a depot at two degrees, sits forty minutes on a dock in July, and arrives with a temperature record that looks perfect. The record is not lying; the tag was simply measuring the air inside a box that never warmed. Cold-chain sensing fails more often through placement and logic than through hardware.

What a temperature tag actually measures

The sensor reports the temperature of the sensor, and nothing else.

A passive tag with a sensing element harvests its power from the reader field, so it can only report when interrogated.
A battery-assisted tag samples on a schedule and keeps a history, which is what makes a journey auditable.
The measured value is the microclimate at the label, not the product core, and the difference can be several degrees.
Response time matters: a sensor with a slow thermal mass will under-report a short excursion.
Accuracy claims usually apply to a range, and the error at minus twenty is not the error at plus twenty.

Passive, semi-passive and active: choose by chain length

The three families are not competitors; they answer different questions.

Passive sensing tags are for spot checks at handover: fast, cheap, no history.
Battery-assisted passive (BAP) tags log continuously and are read when a reader is in range, giving history without infrastructure.
Active tags with mesh or cellular backhaul report in near real time, at a cost that suits high-value or regulatory cargo.
For a four-hour last-mile delivery, continuous logging matters more than real-time reporting.
For multi-day international shipment, the ability to prove the whole journey to an auditor matters more than either.

Reading history without opening the box

A logger that requires a handheld at every step will not be read at every step.

Fixed portals at dock doors read tags as pallets pass, capturing a timestamped checkpoint automatically.
Handheld reads at handover give a manual checkpoint, and are the realistic option for small depots.
Some tags support a printable summary, so the receiver can see the excursion before accepting the goods.
Data must be extractable without destroying the label, or the physical evidence disappears at the first inspection.
Plan the read points first, then choose the tag; the reverse produces a beautiful log nobody collects.

Placement decides whether the record is true

Most disputed cold-chain records are placement problems, not sensor problems.

Place the tag in the same thermal environment as the product, not against the box wall nearest the door.
For palletised goods, place at the core and at the edge; the pair tells you whether the pallet was actually equilibrated.
Avoid placing the tag against pre-chilled gel packs, which will make it read colder than the product.
In a refrigerated truck, the return-air sensor and the product sensor disagree by design; know which one your compliance regime accepts.
Fix the placement in a work instruction with a photograph, because "inside the carton" is not a specification.

Calibration, accuracy and the audit trail

A number without a calibration history is an opinion.

Buy tags with a calibration certificate traceable to a national standard, and keep the certificate with the batch record.
Recalibrate on the manufacturer's interval; a two-year-old logger in a regulated chain is a finding waiting to happen.
Record the tag identifier in the shipment record, so the log can be tied to a physical device.
Store raw data, not only the summary, because auditors ask for the series behind the graph.
Protect the audit trail: signed exports, immutable storage, and a defined retention period.

Excursions: thresholds, duration and MKT

The interesting question is never whether the temperature left the range, but by how much and for how long.

Define the acceptable range per product, not per warehouse, because the same truck carries products with different limits.
Record excursion duration, peak and the area above the limit; a two-minute spike and a two-hour drift are different events.
Mean kinetic temperature condenses a variable profile into one number and is the language regulators accept for some products.
Some products tolerate a cumulative budget better than a single peak, so the logic must match the stability data.
Automate the decision: pass, quarantine, or refer to quality, with the rule written down and version-controlled.

From log to batch record

Sensing only pays off when the reading changes what the system does.

Write the tag identifier onto the receipt line in Odoo, next to the lot and the expiry.
Attach the excursion flag to the lot, so a downstream pick inherits the restriction automatically.
Block put-away for a lot in quarantine, rather than relying on a human to remember.
Keep the raw series attached to the stock move, not to a shared folder with an ambiguous name.
Report on excursion rate by lane and by carrier, because that is where the money is lost.

Economics and where the technology stops

Temperature RFID is not free, and pretending otherwise produces abandoned pilots.

A logging tag costs many times a passive label; use it where the value or the regulation justifies it.
Portal infrastructure is a fixed cost per dock, so the case improves with volume per lane.
For short domestic routes with stable products, a passive check at dispatch and receipt is often enough.
Sensors cannot fix a broken cold room; if the compressor is failing, the data will only document the loss.
The realistic goal is a defensible record and a fast quarantine decision, not a perfect temperature everywhere.

Deployed with that goal in mind, a sensing tag turns a verbal assurance at the loading dock into evidence a customer, an auditor and an insurer can all read the same way.

The teams that get this right treat the temperature record as a document with an owner, a version and a retention rule, not as a by-product of a shipment. They also accept a comfortable truth: you will never know the temperature of every box, so the design goal is to know the temperature of the boxes that would hurt you, and to know it early enough to act.