CMMS reporting: which maintenance reports to run, who reads them, and how to keep them honest

Start from decisions, not from the report menu

Most maintenance systems offer far more reports than any team can read. The usual mistake is to schedule a lot of them, email them to everyone and hope something useful comes out. Nothing does. Work the other way round. Write down who has to decide something, what they need to know, and when they decide it. Then pick the one report that answers that question. If nobody would act differently after reading a report, you don't need it.

The weekly pack: three reports, one meeting

The weekly reports are for running the work, not for judging it. Keep them short enough to cover in a fifteen minute stand up with the supervisor and the lead technicians. Three reports are enough:

1. Open work orders, sorted by priority and then by age in days, not by work order number. Add a column for the reason each one is waiting: parts, a contractor, a production window or a person. A job waiting on parts calls for a different fix than a job nobody has picked up.
2. Preventive maintenance due and overdue, covering the next seven days plus everything already past its window. Show the original due date as well as any new one, so you can see when a job has been pushed back more than once.
3. Repeat breakdowns: any asset with two or more breakdowns carrying the same failure code in the last 30 days. This is the list where root cause work starts. Without it, the team keeps doing the same quick fix every Tuesday.

The meeting needs a fixed rhythm or it falls apart. Pull the reports on Friday afternoon, while that week's jobs are still fresh in people's minds. Use that moment to fill in missing codes and hours. Review the reports on Monday. Every item that needs action leaves the meeting with one owner and a date. The following week's pack then shows whether those actions actually happened.

Monthly and quarterly: trends, not lists

A single month's figure tells you very little. One bad week, a long shutdown or a public holiday can move it a long way. Monthly reports should show at least the last six months next to each other, so the manager reacts to the direction the numbers are moving and not to one month's noise. These are the five numbers worth trending:

• PM compliance: preventive jobs completed inside their window, divided by jobs that fell due in the month. Report deferred and cancelled jobs as separate numbers. Don't quietly leave them out.
• Backlog in crew weeks: estimated hours of open work divided by the hours your team can actually spend on maintenance in a week. A figure like 'three weeks of work waiting' tells you more than '212 open orders'.
• Downtime by asset and cause, ranked from most to least, so the top two or three assets stand out.
• Cost by asset, with internal labour (hours times an hourly rate) next to parts and contractor invoices.
• Planned versus reactive share of labour hours. If the reactive share is going up, planned work is losing the competition for people.

The quarterly review is a different kind of document. It is for capital and contract decisions, not for daily running. Take the five assets that cost the most or stopped the longest over the last twelve months. Put their full history on one page each: failures by code, downtime, total maintenance cost and the age of the asset. That page turns the question 'should we replace it' into something you can answer with evidence and not only with opinion.

Eight traps that make maintenance reports lie

Most bad decisions made from a CMMS don't come from a broken report. They come from a correct calculation run on data that means something different from what the reader assumes. These are the ones to look for:

• Closed date is not completion date. If technicians close work orders in a batch at the end of the month, repair times look long and every job seems to land in the last week. Report on the date the work was actually finished. If your system only stores the close date, make 'close it the day you finish it' a firm rule.
• Rescheduling resets the clock. If moving a preventive job's due date counts it as on time against the new date, compliance can read 100 percent while real intervals stretch out by weeks. Measure compliance against the original due date.
• A count hides what was skipped. Compliance of 95 out of 100 sounds fine, unless the five missed jobs were the boiler and the fire pump. Next to the percentage, list the missed jobs on critical assets by name.
• The 'other' code swallows the analysis. If 'other' or 'general fault' is the biggest bar on your cause report, the report is only telling you how people fill in forms. Read those work orders, group them and add the codes that are missing.
• Work order duration is not downtime. A repair can take two hours of hands-on work while the machine stands for two days waiting for a bearing. Record the time the machine stopped and the time it restarted separately from labour hours.
• Missing internal labour makes in house repairs look free. A cost report built only from parts and invoices favours keeping old equipment running, because the dozens of hours your own people spend on it never appear.
• Stale meter readings break hour based schedules. If PM is due every 500 running hours and nobody has entered a reading for three weeks, the overdue report says everything is fine. Add 'last reading date' to the weekly pack.
• One average across different machines means nothing. A single mean time to repair for forklifts, presses and air handlers mixes jobs that are not alike. Split reliability figures by asset type before you compare months.

A ten minute data check before any report goes out

You can't fix data quality with one big cleanup. You fix it with a small check done every time, by the person who sends the report, in the same order every time:

1. Filter work orders closed this period with zero labour hours or an empty completion note, and send them back to the technician.
2. Filter breakdowns with no failure code, or coded 'other', and code them while people still remember the job.
3. List open work orders with no update in the last two weeks. Either they are really waiting, with a reason written down, or they should be closed or cancelled.
4. Check that every asset with an hour or kilometre based schedule has a reading from the last week.
5. Check that every cost line belongs to an asset. Costs without an asset disappear from cost by asset reports.

Example: what a one page monthly report looks like

This is an illustrative example, not a real site. Picture a small plant with about forty assets. The monthly page has three parts. At the top are four trend lines: PM compliance against original due dates, backlog in crew weeks, total downtime hours, and the planned share of labour. In the middle is downtime by asset for the month, ranked. At the bottom are three actions, each with an owner and a date.

In this example the plant lost 41 hours of downtime, and two assets account for 29 of them. That is where the three actions go: a root cause review of the repeat jam on packing line 2, a spare valve kit held in stock for compressor 2 because most of its hours were spent waiting for parts, and an inspection of forklift 3's charging routine. The other assets wait their turn. Ranking turns a pile of numbers into a short list of things to do. If your work orders, breakdowns and costs are recorded against the same asset, for example in Solmex, this page comes from one set of records and you don't have to stitch it together from three spreadsheets.

Frequently asked questions