Replacing paper temperature logs with automatic monitoring

The problem

A paper log records what someone wrote down when they remembered, not what the fridge did. Between the last check on an evening shift and the first check the next morning there is typically a ten to fourteen hour hole, which is exactly the window in which a door gets left ajar or a compressor gives up. The second problem is quieter: because entries are made by hand, they can be completed in a batch at the end of a week, and everyone in the building knows it. When an auditor or an insurer asks for evidence, a sheet of identical handwriting is worse than nothing.

What you actually need to replace

Before buying anything, be clear which of three jobs the paper log is doing, because they need different answers.

Most sites find that once alerting exists, the recording problem solves itself — the record is a by-product of the monitoring rather than a separate chore.

• Detection — knowing that something has gone wrong while it can still be fixed. Paper does this badly; this is the part sensors fix outright.
• Record keeping — being able to show what the temperature was on a given date. Paper does this adequately until it is questioned.
• Accountability — showing that someone saw the problem and acted. This is the part people forget to specify, and it is why acknowledgement matters as much as alerting.

Does an automatic log satisfy an auditor?

In our experience auditors are more comfortable with continuous digital records than with handwritten ones, provided three things are true: readings are timestamped and cannot be edited after the fact, the export covers the period requested without you rebuilding it manually, and there is evidence of what happened when a reading went out of range.

That last point is where digital systems are sometimes weaker than paper. A dashboard that shows a two-hour excursion with no note attached invites the question "and then what?". Systems that let a manager acknowledge an alert and record the action taken close that loop. We set alerting up this way by default on temperature monitoring deployments.

Check your own scheme requirements — HACCP-based food safety plans, pharmacy standards and customer contracts all word this differently — but none of the UK schemes we work with require the record to be handwritten.

Choosing sensors for the environment, not the spec sheet

The commonest reason a monitoring project fails is not sensor accuracy. It is that the sensor cannot get its readings out of the space it is in.

Insulated walls, metal shelving and the door of a walk-in are effective at blocking consumer Wi-Fi. On multi-site jobs we survey coverage before quoting sensors, and it is not unusual for the first piece of work to be network remediation rather than monitoring — that is what happened in the cold chain monitoring case study.

• Walk-in chambers and freezers — probe sensors with the transmitter outside the cold space, or long-range wireless that tolerates the enclosure. See cold storage monitoring.
• Under-counter and display fridges — self-contained wireless units, one per appliance, positioned away from the fan outlet. See commercial fridge and freezer monitoring.
• Zoned ambient storage — fewer sensors, placed to represent the zone rather than one shelf. See warehouse temperature monitoring.
• Medical and vaccine storage — calibrated probes in glycol or a buffered solution so short door openings do not fire alerts. See vaccine and pharmacy fridge monitoring.

Set thresholds with a duration, not just a number

A fridge that reads 6°C for ninety seconds during a delivery is behaving normally. A fridge that reads 6°C for forty minutes is failing. If your alert rule only knows the number, staff will receive alerts every time someone opens a door, and within two weeks they will mute them. That is the single most common way a monitoring rollout quietly dies.

Set a threshold and a duration for every sensor, then tune both during the first month using the real data rather than guessing up front. Expect to adjust: the first set of rules is a starting position, not a configuration.

How to phase the changeover

Do not switch off paper on day one. Run both for a period — typically four to six weeks — so you can compare, prove the digital record is complete, and get staff used to acting on alerts while the fallback still exists.

A workable sequence for a single site:

• Survey coverage and pick sensor placement per appliance or zone
• Install and let the system record for a week without alerting anyone, to learn normal behaviour
• Turn on alerting with generous durations, routed to named people rather than a shared inbox
• Run parallel with paper for a month, tightening thresholds as the noise becomes obvious
• Retire paper for monitored equipment only, and write down who checks the dashboard and how often
• Review after a quarter: which alerts fired, which were real, and which sensors need moving

What it costs to run, honestly

The purchase is the small part. Batteries need replacing, sensors get knocked off mounts, and someone has to own the alert list when a person leaves. Budget for a light ongoing responsibility rather than a one-off install, and decide early whether that sits with you or with us as part of IoT delivery and support.

The cost comparison that usually matters is not sensors versus clipboards. It is sensors versus one avoided stock loss, or one contract retained because you could produce the records.