What is Maintenance?

Why the obvious answer isn't enough

Most people describe maintenance as preserving machinery and fixing it when it fails. That view centres on the asset and the maintenance management system sitting behind it. But machinery has changed enormously in recent decades — from analogue to digital, full-authority control, sensing, and automated protection — to the point that even a car can no longer be maintained without electronics.

At the same time, the range of stakeholders with real power and influence over how machinery is used and maintained has widened. It's no longer just internal teams and customers: regulators, government, media, communities, and increasingly the environment itself all carry expectations around safety, sustainability and social responsibility. Meeting that wider set of expectations means maintenance has to be treated as strategic, and as a core part of achieving operational excellence — not an isolated technical function.

A stakeholder-based definition

A useful starting point is Moubray's definition of the purpose of maintenance:

Swapping "preserving machinery" for "preserving functions" sounds like a small change of wording, but it has profound implications. Take a simple example: what is the primary functional requirement of an F1® racing car while it's competing? Most people would say "to win races." But if that same car later becomes a static exhibit in a museum, its function shifts entirely — now it exists to leverage its racing history and draw visitors to the brand. It's the same physical car, but the maintenance needed to preserve its function is completely different depending on context.

A wider perspective on what counts as maintenance

Once you think in terms of stakeholder functions, activities that were never logged in a CMMS start to look like maintenance too. A pilot's pre-flight walk-around is a functional safety check — and it is maintenance, even though no one would typically label it that way.

Defect elimination is another example: using historical failure data and root cause analysis (or a risk-based view where data is sparse) to find and correct common causes of failure, whether that means fixing a process or modifying the machine itself. So is predictive maintenance — using time-series and condition data from asset sensors to diagnose incipient failure and estimate remaining useful life before a breakdown occurs. Increasingly, that analysis should be automated using computing and machine learning, rather than run manually.

The common thread: a machine doesn't have to be broken to be functionally failed, not all maintenance work involves a spanner, and not all of it lives inside a maintenance management system. In many organisations, related work is also split across organisational boundaries — planned maintenance sitting apart from scheduled vibration analysis, for instance — and it's worth asking whether that output data is ever merged, or whether the tasks are rationalised between the two functions at all.

Why this matters for how you plan

Understanding the true purpose and scope of maintenance is the foundation for thinking clearly about its effectiveness and efficiency. Once functions, not just assets, are the starting point, it becomes far easier to see where preventative, predictive and defect-elimination work actually overlaps — and where it's quietly being duplicated or missed across organisational lines.