What to ask a CMMS for predictive maintenance

If you’re looking for a system with the intention of doing predictive maintenance, it’s worth knowing what to ask. “We do predictive” shows up on almost every spec sheet and means very different things: from storing a text field with the reading to automatically generating work when a value crosses a line.

This is the list of what to demand, with what GMAO CLOUD does on each point.

1. That the asset can have a counter

The basics, and not everyone has this properly solved. The equipment needs its own counter — running hours, mileage, cycles, units produced — not a loose field where you jot down a number.

In asset management the counter is an entity with its own unit, and every reading gets logged with who took it and when. That’s what lets you later have a series instead of a snapshot.

2. That a threshold generates work, not just an alert

This is where the difference lies between a system that records and one that acts, and it’s the question that separates products the most.

In GMAO CLOUD, on an asset’s counter you define a limit — the units it’s expected to hold up to — and a warning percentage. When a reading gets logged, the system adds up the accumulated total, calculates the percentage consumed, and, if it exceeds the threshold, automatically generates the preventive work order, with its asset, address, and the checklist template that corresponds to that equipment.

In other words: the plan doesn’t depend only on the calendar. It can depend on actual usage, which is what’s needed for a machine running irregular shifts, a fleet, or a component whose wear runs by cycles.

Ask this literally in any demo: if a reading crosses the threshold, what exactly happens? If the answer is “it shows up in a list,” that’s not the same thing.

3. That checks accept values, not checkboxes

A checklist of checkboxes says someone looked. One with a minimum and maximum value says what they saw, and logs any out-of-range reading as an anomaly the moment it’s taken.

This is what turns an inspection round into a time series. And here’s the most useful idea in the whole matter: you don’t need sensors to start doing predictive maintenance. Someone with a thermometer and a portable vibration meter, going through the critical equipment every month and jotting down values, already generates the trend that lets you anticipate.

Continuous sensing speeds that up. It doesn’t invent it.

4. That checklists can be defined by family

If you have to configure the round equipment by equipment, nobody does it on a site with fifty identical pumps.

Checklist templates resolve in cascade — asset, model, subfamily, family — so you define it once for a whole type of equipment and only fine-tune where needed.

5. That it can be recorded offline

Measurements get taken in plant rooms, pits, and warehouses, which is where there’s no signal. If the technician has to write it on paper and enter it in the afternoon, the values come out rounded and the series loses precision exactly where it matters.

The technicians’ app stores orders, assets, and documents on the device and queues actions taken offline, syncing them once signal is back. If one fails, it stays marked with its reason instead of disappearing.

6. That the anomaly turns into something

The point where most predictive projects die, and it isn’t technical.

A technician logs a vibration that’s gone up, and nothing happens. By the third time, they stop logging it, and rightly so. The anomaly has to be able to turn into an incident with its priority and owner, or into an explicit decision to do nothing.

And notifications have to be tunable: you can mark which statuses should not notify. Notifying on everything turns notifications into noise, and people stop reading them, which is worse than not having them at all.

7. That there’s history to compare against

A threshold doesn’t come from a catalog: it comes from the behavior of that piece of equipment at that site.

From there follows something not entirely intuitive: predictive maintenance starts by measuring equipment that’s working fine, to establish a baseline. If you only measure when you suspect something, you’ll never have anything to compare against.

That’s why the real requirement, ahead of any predictive feature, is that work orders get closed with accurate data: times measured with a stopwatch, material logged, and checks recorded. Predictive maintenance on an empty history doesn’t exist.

8. That accumulated cost also triggers something

There’s a variable that isn’t physical and that decides more money than vibration does: how much you’ve spent repairing that machine.

On an asset you can record its replacement cost and a warning percentage: the system notifies when accumulated repair spend exceeds that percentage. It doesn’t generate an order — replacing a piece of equipment is a business decision, not a task — but it puts the figure in front of you exactly when it’s time to look at it.

9. That the data can get out

If your medium-term plan includes your own analytics — cross-referencing maintenance with production, or applying your own models — you need the history to be exportable from the system.

GMAO CLOUD has a public REST API and direct SQL access, the two paths BI tools ask for. They’re listed in the integrations catalog alongside SOAP, CSV, and FTP.

It’s the honest answer to “do you have custom dashboards?”: inside the product there’s a catalog of reports with filters and export, and for anything beyond that, your own tool connects, which also lets you cross-reference with data the CMMS doesn’t have.

10. That nobody sells you more than what’s there

One final warning. A CMMS is where the history and the decision live: the asset record, the readings, the anomalies, the order that comes out of them, and the indicators that tell you whether it worked.

It isn’t a continuous signal-analysis platform, and whoever sells it to you as one is probably describing something else. If your case demands high-frequency sampling on the raw signal, that’s a separate system that connects — via API — to the CMMS so the alert ends up becoming work.

How to start without spending

Three or four critical pieces of equipment. For each: what failure mode you want to anticipate, what variable reveals it, how often it’s measured, and who acts when the threshold is crossed. The round gets set up as preventive maintenance with its value-based checklist, and within a few months you have a baseline.

There’s more on the method in predictive maintenance: how to get started. If you want to see it on your own equipment, you can request a demo.