Break-fix is a cost centre. Preventive maintenance is a business.

A maintenance plan defines the maintenance schedule for one or more assets and generates work orders for future visits in advance. Assets are attached to the plan through maintenance asset records, so a single plan can cover many machines, and maintenance work rules set the frequency — either calendar-based or driven by usage and meter readings. Generation runs ahead of time, which is what puts planned work on the dispatch board months before it is due and makes capacity planning possible rather than reactive.

With the assets under contract, not with everything you have ever installed. Trying to build a complete historical register before starting is how these programmes stall for a year. Capture the equipment covered by live agreements first, because that is where the revenue and the obligation already exist, then let technicians create and correct asset records as they visit sites. Within a couple of service cycles the register becomes genuinely representative, and it has been built by the people who can actually see the machine rather than by a data project.

As deep as the level you replace or report on, and no deeper. If you swap compressors and want to know which model fails soonest, the compressor needs to be an asset. If you never track a component individually and would never quote for it separately, it does not. Excessive depth is a real cost, because every level has to be kept accurate as equipment is replaced, and a hierarchy nobody maintains is worse than a flat model. The practical test is whether a service manager would ever ask a question that requires that level to answer.

It depends on the asset class, and most fleets need both. Calendar-based intervals are simple to sell, simple to plan capacity around, and mandatory where regulation or the contract sets the frequency — but they over-service lightly used equipment and under-service heavily used equipment. Usage-based intervals match the real wear on the machine and avoid paying for unnecessary visits, but they depend entirely on meter readings arriving reliably from technicians, connected equipment or the customer. If the readings will not arrive consistently, a calendar schedule you can actually honour beats a usage schedule you cannot.

By making coverage visible on the device before the technician commits to anything. Entitlements and warranty terms attached to the asset and the contract can drive what the technician sees on the work order — whether this visit is covered, chargeable, or outside the agreement entirely. The revenue leak in most service businesses is not deliberate; it is a technician on site making a reasonable judgement without the commercial information, and nobody reconciling it until the quarter closes. Surfacing entitlement at the point of work is one of the highest-return configurations in the whole platform.

Condition-based maintenance is realistic today and genuinely valuable: connected equipment reports a developing fault, and a work order is raised against that specific asset before it fails. True prediction — forecasting failure from historical patterns — requires a volume and quality of service history that most organizations do not yet have, precisely because their asset records are thin. The honest sequence is to build the installed base, capture consistent service history and failure causes, connect condition data where the asset is critical enough to justify it, and treat prediction as something that becomes available later rather than something you buy now.