What to Write Down So a Reading Survives

Why this matters

A reading has two lives. The first one lasts about ten minutes: the tech takes it, decides something, and acts. The second one starts months later, when a warranty claim, a callback, a dispute or a sale forces someone to open the file and ask what the machine was doing back then. Almost every reading a shop records survives the first life and dies in the second, and it dies the same way every time: the number is there and the conditions are not.

The rule that sorts a good record from a useless one is narrower than most forms suggest. Write down what cannot be recovered later. Everything else is noise, and noise on a record is not free, because it gives a second place for an error to live and it buries the fields that are load-bearing.

The test the record has to pass

Hand it to somebody who was not there, months from now, and ask them to re-decide. Not to re-read your conclusion. To reach a conclusion themselves from what is written.

That is a harder bar than it sounds, and it is the correct one, because the second life of a reading is always adversarial in some small way. The manufacturer wants to know whether it was commissioned correctly. The customer wants to know why they paid for the same visit twice. Your own tech wants to know whether the number moved or whether the last guy measured somewhere else. Each of those is a stranger re-deciding, and each of them gets nothing from "checked, within range."

The fields that cannot be recovered later

This is the artifact. Everything on it shares one property: once the truck pulls away, it is gone.

The quantity, with its units and its reference. Pressure with no statement of gauge or absolute is two different numbers. A temperature difference with no statement of which two points is not retakeable. Resistance measured with the circuit isolated is a different quantity from resistance measured in place.

The exact point. Not the equipment, the point. Which port, which terminal, which conductor, which side of which component, and where relative to anything that branches. The single most common cause of two readings disagreeing is that they were taken at two places.

The operating state and the elapsed time in it. Running or off, which stage or speed, how long it had been in that state, measured from a named event. Clock time is not elapsed time and the ticket already has clock time.

The conditions the physics says the number depends on. Ambient, entering conditions, load, filter or screen state, supply voltage. Pick these from the mechanism, not from a generic list: the question is which quantities would move this reading if they moved.

The instrument, and when it was last checked. Identity and check date. This is what lets a later reader decide whether a difference between two readings is the machine changing or two instruments disagreeing, and it is the field that makes a trend defensible.

The raw indication, before any correction. If you know your instrument's offset and you correct for it, record both the indicated value and the corrected one, and say which is which in the same line. A record holding only corrected values cannot be re-corrected when the instrument's offset is later found to have changed.

Who took it. Not for blame. Because six months on, the fastest route to a missing field is asking the person who was standing there, and a record with no name has no route.

What to leave off, and why each one costs you

Four things get written on service records constantly and each of them makes the record worse.

Values you derived on site. A difference, a percentage, a rise, a projection. Record the two raw numbers and let the difference be computed by whoever needs it. A derived value in the record is a place for an arithmetic error to become permanent, and it hides the inputs that a later reader might need for a different comparison than the one you made.

Judgments in place of numbers. "Normal," "within range," "a little high." These are the conclusion, not the evidence, and the conclusion is exactly the thing the second reader is trying to check. Write the number and the judgment, in that order, and never the judgment alone.

Numbers rounded to the decision. Rounding 10.4 to 10 because the threshold is 12 destroys the only information a trend could have used. The decision can round; the record does not. Related and separate: the last digit an instrument displays is not necessarily a digit it can guarantee, and a sibling article covers that distinction, so record what the instrument showed and let the accuracy question be settled by the instrument's stated specification rather than by rounding in advance.

Anything a photograph carries better. A nameplate, a wiring configuration, a corrosion pattern, a fitting arrangement. Transcribing a nameplate by hand introduces errors and takes longer than photographing it. Photograph it and record which piece of equipment it belongs to, because a photo with no anchor is as orphaned as a number with no point.

Units and reference: the field everyone thinks is obvious

It is obvious at the moment of writing and it is the field that most often kills a record later, because the writer and the reader are the same person on the same day and no ambiguity is visible.

A pressure that does not say gauge or absolute. A temperature difference with no direction. A current with no indication of whether it was a running value, a starting value or a maximum captured over some window. A flow with no statement of the conditions it is referred to. In each of these the number is fine and the record is unusable, and none of them can be repaired without going back.

The fix is a form field, not discipline. Discipline fails at the end of a long day. A field that will not close without a unit and a reference does not.

How long it has to last, and one regime that is not yours to set

For equipment readings, the useful life of a record is set by the questions that will be asked of it: the warranty term, the service agreement term, the equipment's life, and any dispute window your own attorney tells you to plan for. Those are business decisions and they belong in the shop's own retention policy.

One category is not a business decision. Where a reading is an employee exposure record for a toxic substance or harmful physical agent - air sampling, noise dosimetry and the like - 29 CFR 1910.1020 requires the employer to preserve it for at least 30 years, and employee medical records for the duration of employment plus 30 years, with employee access rights attached. That regime is triggered by what was measured and who was exposed, not by what instrument was used, so an ordinary equipment reading does not fall under it, and a noise or air measurement taken on your own crew does. Know which one you are holding before you set a purge date.

Worked example: the same reading, written two ways

A commissioning visit on a light commercial air handler. The tech measures external static pressure and gets an indicated 0.9 inches of water column, above what the unit's documentation gives for its rated airflow. He notes it, adjusts nothing, and leaves.

Version A, as most records are written: "ESP 0.9 - high, advised customer."

Version B, on a form with the fields above: external static pressure, indicated 0.9 in. w.c., manometer identified with its last check date, measured across the unit at the two factory test ports, unit running in high-stage cooling for 14 minutes at the time of reading, filter new with its type and nominal rating noted, all supply registers open, outdoor temperature recorded, entering air temperature recorded, no correction applied because the instrument's last check found no offset.

Fourteen months later the blower motor fails and the manufacturer asks whether the unit was commissioned within its stated static limits.

Version A can answer nothing. It cannot even establish whether 0.9 was measured across the unit or across the whole system, which are different quantities with different limits. The claim gets argued on goodwill, and the shop has no position because it cannot show what it did.

Version B answers directly. The two test ports fix the quantity as external static across the unit. The filter state removes the most common alternative explanation. The 14-minute runtime establishes the unit was in a settled operating state rather than mid-ramp. The instrument's check date closes off the "your gauge was wrong" line. The record shows a static above the documented value at commissioning, which either supports a claim of a duct system problem the shop flagged and the customer declined, or it does not, and either way the shop is arguing about facts instead of about memory.

The cost difference at the time of writing. Version B is roughly two additional minutes of writing on a visit already measured in hours. The recovery cost of Version A is a return trip that cannot recover the commissioning conditions at all, because the filter has been changed several times since and the unit's operating state fourteen months ago is not a thing anyone can go and measure. That asymmetry, seconds against an unrecoverable, is the entire argument for the form.

How to verify your form is good enough

Do not audit the records. Audit the questions.

Pull the last handful of arguments the shop had - a warranty claim, a callback, a customer dispute, a disagreement between two techs - and for each one, write down the question that had to be answered and check whether the record answered it. Every question that had to be settled by a phone call to the tech who was there is a missing field, and it is missing from every record that tech takes, not just from that one.

Then fix the form rather than the record. A missing field found this way appears on hundreds of past jobs and cannot be added to any of them, which is unpleasant, and it can be added to every future one in about five minutes of form editing, which is the only part still available to you.

References

• 29 CFR 1910.1020, OSHA general industry, access to and retention of employee exposure and medical records, including the 30-year preservation requirement for exposure records
• Manufacturer commissioning documentation, which states the values, the measurement points and the conditions a claim will be judged against
• Trade-standard practice for commissioning reports and service documentation
• See related: Why the Last Digit Is Not the Measurement; What a Baseline Is Worth and When to Take It