Maintenance report: turning visit records into decisions

A maintenance report is the periodic summary of what the maintenance function did and what it found — not a printout of every job. Its purpose is to support three decisions: what to repair, what to replace, and where to spend next year. A report that lists two hundred closed work orders without answering any of those has confused activity with information.

Maintenance report: the four views worth producing

• Work completed against work raised, so the queue trend is visible rather than implied.
• Reactive versus planned split, by count and by hours — the clearest measure of whether the function is ahead of failure or behind it.
• Cost by asset, with parts and labour separated, ranked worst first.
• Repeat failures: assets attended more than twice in the period, with the same or a related fault.

Four views, and the fourth is the one that changes budgets. An asset visited five times in a quarter has already cost more than its replacement in a way that no single invoice reveals.

What the data has to look like

1. Every job attached to an asset from a controlled list — free-text locations make aggregation impossible.
2. Job type recorded as planned or reactive at the point of raising, not guessed afterwards.
3. Labour hours captured on the job, even approximately, because a report without hours can only discuss parts.
4. Parts recorded with cost, or at least with part numbers that can be costed later.
5. Fault cause coded against a short list — a list of thirty causes gets used as free text.
6. Downtime recorded where the asset matters to production, separately from labour time.

Approximate hours consistently recorded beat exact hours recorded half the time. A report is a comparison over periods, and a gap in the data is worse than a rounding error.

Where the numbers come from

If the jobs live in a register with asset, type, hours and cost as real columns, the maintenance report is a set of filters rather than a monthly assembly exercise — Ettex Records produces those views from the same rows the team already closes, and the underlying readings and parts arrive from the field through Ettex Forms. The alternative, which most small maintenance functions live with, is a spreadsheet rebuilt each month from memory and email, which is why the report is usually late and never trusted.

Writing it so it gets read

• One page of findings, with the tables behind it as an appendix.
• Lead with the decisions requested, not with the volume of work done.
• Name the three worst assets and what you propose for each.
• Show the planned-versus-reactive trend over at least six periods — a single month means nothing.
• State what you could not measure, so the gap gets funded rather than forgotten.

The report draws on two upstream habits: work order tracking supplies the closed jobs with their states and hours, and planned preventive maintenance supplies the planned side of the split. Where an asset register exists, the report can finally answer questions per asset rather than per month.