Documenting maintenance work

Documenting work has a bad reputation among the people who do it, and for good reason: for years it meant filling out a piece of paper at the end of the shift so someone in the office could type it in again. Seen that way, it’s pure cost.

The way to stop it being that isn’t insisting on documenting more. It’s changing when the record happens and what gets recorded.

What makes a record valid

Four elements. If one is missing, what’s left is a note, not evidence.

Date. When it was done, not when it was written down. These are two different things, and the difference matters when you need to prove something.

Author. Who did it, identified. There’s a detail here that breaks everything else if it fails: if several people share a login, the record doesn’t identify anyone. That’s what happens when registering someone costs money; in GMAO Cloud licenses are unlimited on all three plans precisely so every job stays attributed to whoever did it.

What was done or checked. Not “inspection carried out,” but the specific points and, when there are any, their values.

Sign-off. The signature of whoever receives the work, when the workflow requires it.

What needs to be recorded for every job

On the work order itself, without leaving it:

Time, measured. With a stopwatch inside the order itself, not estimated at the end of the day. Remembered hours always come in low, and with them the job’s cost is a guess. The order also stores the target time, the estimated time and the deviation between planned and actual, which is the data point that lets you quote better next time.

Material consumed, against the warehouse and with its batch when applicable. Not as a later entry: as a consequence of closing the order.

Checks, with values. This is the difference between a record that’s useful and one that only proves someone showed up. Checklists support fields with a minimum and maximum value, so a reading out of range gets logged as an anomaly on the spot instead of as a side comment no one will ever re-read.

Photos, before and after. They stop staying stuck on the phone’s camera roll.

The cause, when there was downtime. The order stores the downtime cause, which is what later lets you analyze why something keeps repeating.

What’s left pending. The order can be flagged as pending material and as needing a return visit. Without those statuses, that work lives in a notebook and the client remembers it before we do.

Where it has to happen

At the facility, not in the office. This is the change that makes everything above possible.

The technician app stores orders, assets and documents on the device itself, and every action taken offline —changing a status, logging hours, consuming material, filling in the checklist, collecting the signature— goes into a queue that empties once signal returns. If one fails, it doesn’t disappear: it stays flagged with its reason.

This isn’t a convenience: maintenance happens in basements, warehouses, plant rooms and elevators — exactly where there’s no signal. An app that requires a connection records nothing right where the work is heaviest.

How not to add work for the technician

The practical rule: every field you ask for has to come back to them as something useful.

The equipment’s history and documentation on the phone, so they don’t have to call the office. The anomalies left open last time, so they don’t repeat the diagnosis. The material they’re carrying in the van, so they don’t have to go back for a part.

And the reverse: if a field serves no purpose, remove it. Multiplied across every order in the year, one useless field is a lot of time and a fair amount of resistance.

The other rule, even more important: what gets recorded has to have a consequence. If they detect an anomaly and it doesn’t turn into an issue with an owner and a date, by the third time they’ll stop logging it. And they’ll be right.

The documentation that goes with it

Beyond what gets done, there’s what describes the equipment: manuals, drawings, certificates, data sheets, inspection reports.

Document management hangs them off the asset, the client, the address, the order or the issue —and almost any other entity— with its visibility set (who sees it: client, technician, provider) and its expiry date when it has one, with a daily check on what’s about to expire.

Certificates also carry their number, scope, issuer and holder, because a certificate isn’t just any PDF: it’s proof that something was done in compliance with something.

What it’s useful for afterward

Three uses, and the order they appear in isn’t the order they’re valued in.

Operational. A technician who arrives at a piece of equipment and sees what was done to it last time fixes it instead of patching it again.

Financial. With measured time and billed material, reports give cost per asset, hours per client and the deviation between estimated and actual. That last one usually reveals a type of job being sold below what it costs.

Defensive. The day there’s a claim, an incident or an inspection, what gets asked for is the record. Legal maintenance rests exactly on this: order history, completed checklists and documentation with its dates. The system records and demonstrates; the one who complies with the regulation is the company.

What makes the record reliable

Deletion is logical, not physical: what gets deleted by mistake can be recovered.

There’s a record of who created, modified or deleted each thing, and when. In a dispute over whether something was delivered or changed, that separates a fact from a version.

And a user who leaves doesn’t get deleted. Their access is revoked and the account is kept; if it were deleted, every order they closed would lose its author and, with it, its value as evidence.

Where to start

With the minimum that makes a record valid: measured time, consumed material and a completed checklist, on critical assets. Photos, signature and the rest of the fields get added later, once the habit has already formed.

Starting by asking for twenty fields guarantees none of them get filled in properly.

If you’d like to see what would get recorded from one of your own jobs, you can request a demo.