Temperature logs: why they get ticked but tell you nothing

A temperature log is useful in only two cases: when an out-of-range reading triggers a reaction written down in advance, and when the history lets you see drift before the breakdown. A series of numbers ticked every morning, with nobody comparing them and nobody knowing what to do about an abnormal value, protects against nothing.

The same number, every day, for a month

The owner of a restaurant group goes through one site's temperature logs to prepare for an audit. Walk-in cooler: the same value every morning, in identical handwriting, down to the digit. Freezer: no variation. He suspects nobody of fraud: he wonders how a piece of equipment could be so stable. The cook explains that the reading is done in passing, reading the display from memory, or copying the previous day's when the unit is far from the workstation.

What the owner discovers is not an individual fault. The task was never designed properly: the display is far away, the sheet is at the other end, there is nobody to notify when a reading is off, so the log becomes a ritual.

The causes of a worthless log

• An out-of-range reading has no follow-up. If nothing happens when the value leaves the range, the employee learns that entering it has no consequence.
• The limit is not written next to the number. Knowing that a value is "good" requires knowing the limit for that equipment, which is rarely displayed.
• The task is badly placed. The display is in one spot, the sheet in another, and the log comes after prep.
• Nobody reads the history. A slow drift (a door that seals badly, a clogged evaporator) shows up only in the trend, not in an isolated value.
• Measurement and declaration are mixed. On a sheet, a value read from a sensor and a value copied from memory look the same.

The components of a log that works

Element	"Decorative" log	Useful log
Value	A number alone	Number + expected range displayed
Source	Unknown	Measured by sensor or entered by a person, stated
Out-of-range reading	Noted or ignored	Linked corrective action and a named owner
History	In a binder	Searchable by unit and by period
Review	Never	Regular review of the trend

Setting up a log that triggers something

1. List the equipment to monitor. Cold room, freezer, display cases, blast chiller if you have one: one point per unit, not per corner.
2. Set the acceptable range with your food safety plan. It is written on the unit or on the sheet, not in a folder.
3. Put the log where the display is. A tablet or a sheet next to it, or a sensor that sends the reading.
4. Write the corrective action in three lines. Who to notify, what to do with the products, who decides.
5. Distinguish sensor from manual entry. Both are acceptable, but you must be able to tell them apart.
6. Have the trend read every week. Ten minutes: which unit is moving away from its range, which deviation was handled, which is still open.
7. Close the loop. A deviation is closed only when the action is recorded, not when the value returns to range.

Reading a trend without being a technician

You do not need a complicated dashboard to see drift. Look at three things: the average value of a unit this week against last week, the spikes during periods of heavy door opening (deliveries, rush), and the time it takes to get back into range. A cold room that takes longer and longer to come back down is warning you of wear before the breakdown. This reading takes a few minutes if the data is gathered in one place. It becomes impossible if it is scattered across sheets, binders and sites.

Common mistakes

• Multiplying logs to feel reassured. Beyond what is feasible, entry becomes mechanical.
• Treating a missing log as a fault. First find out why the task could not be done.
• Trusting apparent stability. A perfectly flat curve is a signal to check, not proof of good condition.
• Changing the medium without changing the task. A tablet far from the display reproduces the paper problem.

What you can measure

• The delay between an out-of-range value and the first recorded action.
• The number of units whose trend drifts away from the range over several weeks, before any alert.
• The share of logs whose source (sensor or manual entry) can be identified.

Where Tsuno comes in

Tsuno takes in HACCP logs entered by hand or coming from sensors, and keeps each value's source, date and history. Incidents opened after a deviation keep their follow-up, down to the action taken. You can ask Tsuno in plain language which deviations were left without follow-up at a site, and it answers with the elements it relies on, or says the data is missing. It does not decide what happens to the food: people decide and the decision is recorded. Details on the features page.

Further reading

To place these logs in the bigger picture, see digital HACCP: what software should actually remove. For what comes after a deviation, read why restaurant incidents keep repeating despite reports and compliance theater. The opening version is in the opening checklist.

Key takeaway

A log is worth what the reaction it triggers is worth. Put the task next to the display, write the corrective action, distinguish measurement from declaration, and read the trend every week.