Maintenance 4.0: what it is and what it takes

Maintenance 4.0 is the maintenance translation of what got called industry 4.0: equipment that generates data, systems that collect it, and decisions made with that data instead of with habit.

The problem with the term is that it gets used to sell very different things, from a mobile app to an analytics platform. It’s worth separating what’s real, in what order it arrives, and — most important — what needs to be solved beforehand, because 90% of projects that fail do so by skipping the first steps.

The ladder, from the bottom up

Step 1: the record has to exist. Before analyzing anything you need data, and data is born when someone closes a work order with real times, material consumed and checks done. Without this, everything else is a layer built over nothing.

Step 2: recording has to happen where the work happens. If the technician writes it in a notebook and enters it in the afternoon, the data exists but isn’t accurate: hours come out understated and material shows up weeks later.

Step 3: the plan has to execute itself. Preventive maintenance generating orders from the frequency, with its owner and its date.

Step 4: measure with values, not checkboxes. The step almost nobody climbs, and the one that opens the door to everything else.

Step 5: sensor and predict. The one everybody wants to start with.

Why step 4 is the one that matters

A checklist of checkboxes says someone looked. One with values says what they saw.

In GMAO CLOUD, fields support their type, their label and a minimum and maximum value, so a reading outside that range gets recorded as an anomaly on the spot. That turns an inspection round into a time series: that pump’s vibration, that chamber’s temperature, that filter’s differential pressure, measured every month by a person.

And here’s the point that gets overlooked: a time series taken by hand already lets you predict. You don’t need a sensor to see that a value has been getting worse for six months. Sensors speed that up and automate it, but they don’t invent it.

That’s why the order matters. If you don’t have the inspection round with values, installing sensors will just give you a graph nobody knows how to interpret, because there’s no history to compare it against.

What predictive maintenance actually is

Intervening when a measurement indicates degradation, not when the calendar says so and not once it’s already failed. The usual variables are vibration, temperature, power consumption, oil analysis and noise.

It requires three things, and the third is the one that fails:

1. Measuring the right variable for that failure mode.
2. A threshold, which doesn’t come from a catalog — it comes from that specific piece of equipment’s own history.
3. Someone acting when it’s crossed. If the alert doesn’t turn into an incident with its priority and its owner, the whole system stops being worth anything, and the operator stops trusting it.

In GMAO CLOUD this doesn’t stop at recording. An asset can carry a counter — kilometers, hours, cycles — with a limit and a warning percentage; when a reading pushes it past that percentage, the system generates the preventive order automatically, with its asset and its checklist. The plan stops depending solely on the calendar and starts depending on actual usage.

And there’s a second automation, on cost: if the accumulated spend on repairing a machine exceeds the percentage you define over its replacement cost, the system warns you. It doesn’t generate an order — replacing a piece of equipment is a business decision — but it puts the figure in front of you when it’s time to look at it.

Mobility, the 4.0 part that’s already solved

Of everything sold as 4.0, this is what genuinely changes day-to-day work, and it already works.

The technician app stores orders, assets and documents on the device itself and queues every action taken without coverage — changing status, logging hours with a stopwatch, consuming material, filling in the checklist, capturing a signature — syncing them once a connection is available. If one fails, it stays flagged with its reason.

Add the QR code on every asset: the technician scans it and has the record, history and documentation in front of them without calling the office. It’s unglamorous and it’s what saves the most time.

Connected data

The other half of 4.0 is systems talking to each other. A closed job shouldn’t need to be typed in twice to get billed.

The integrations catalog has the scope of each connector — worth checking, because not all of them work in both directions — plus the generic routes that cover whatever isn’t on any list: REST API, SOAP, direct database connectivity and file exchange.

The rule that avoids most problems gets agreed before connecting anything: the ERP owns clients, items and price lists; the CMMS owns assets, interventions and consumption.

And artificial intelligence

Worth discussing without overselling it, because this is where the most hype lives.

In GMAO CLOUD, AI does concrete, bounded things: an assistant that answers in plain language about your assets, your plans and your history — also inside the technician app; document reading, so a supplier’s PDF report turns into data without having to type it in; help drafting the text of a work order; and translation of your own catalogs.

Two things matter more than the list: it proposes, it doesn’t execute — it answers and suggests, it doesn’t close orders or assign work — and every feature has its own on/off switch, with most of them off by default. It’s covered in detail in AI in GMAO CLOUD.

And the usual caveat: an assistant over an empty history knows nothing. Back to step 1.

How to know if you’re making progress

With four figures that come from the reports, not from the list of technology installed:

• Preventive-to-corrective hours ratio. The one that best describes whether the operation is proactive. It takes months to move.
• MTBF per critical asset: reliability.
• MTTR: response capacity.
• Accumulated cost per asset: the one that lets you decide whether a machine gets repaired again or replaced.

If you’ve put sensors on half the plant and these four aren’t moving, the project is technology, not maintenance.

Where to actually start

With steps 1 through 4, on your critical equipment. Checklists by family, frequency, recording in the field, and checklists with values. It’s less impressive than a dashboard full of sensors, and it’s what makes the dashboard mean something once it arrives.

If you want to see how it’s set up, you can request a demo.