The licence renewal lands in your inbox. The CMMS went in three years ago. Half the assets were never loaded. The fitters still write on paper, and the planner types it in on a Friday if there's time.
Pull the breakdown history for your worst machine and you get 400 jobs. Most of them say "fixed". A few say "reset and ran". One says "ask nights".
You paid for a system that was meant to tell you what's failing and why. What you've got is a very expensive diary.
CMMS projects tend to fail in the same place. The software works in the demo. It doesn't fit the way your engineers work at 3am with grease on their hands.
Start with what you need it to do
Before you look at a single system, write down the jobs you need it to do. For most sites the list is short:
- hold every asset, with a parent and child structure that matches the plant
- schedule PPM from the maker's intervals and your own experience
- schedule statutory inspections and hold the certificates
- log breakdowns quickly, with a cause as well as a time
- link spares to the assets they fit
- show you repeat failures, MTTR and the cost of each machine
If a system can't do those well, the extras won't save it. If it can, most of the extras won't matter.
Get the asset list right before you buy anything. Our post on asset registers covers how to build one. A clean register moves from one system to the next. A messy one ruins every system you put it in.
Test it on the shop floor
Here's our view. The demo tells you nothing. The test that matters is your fitter, on nights, on their phone, logging a breakdown.
So ask for a trial and run it on one line for a month. Watch for these:
- How many taps to log a breakdown? If it takes ten minutes at 2am, it won't happen.
- Does it work where your signal doesn't? Plant rooms, cold stores and steel buildings often have none.
- Can someone take a photo of the HMI and attach it to the job?
- Are the failure codes your words, or the vendor's?
- Can a cover engineer log a job without a full licence and a training course?
That last one gets missed. If your nights or weekends are covered by contract engineers, half your breakdowns happen when they're on shift. If they can't log the job, the history has a hole in it every weekend. Our post on inducting a contract engineer covers getting them set up from shift one.
What the law asks your records to do
There's less legal pressure here than vendors suggest, and it's worth knowing exactly what there is.
PUWER doesn't make you keep a maintenance log. The code of practice says so: "There is no requirement for you to keep a maintenance log". It does recommend one for high-risk equipment (L22, paragraph 78). But once a machine has a log, it must be kept up to date (PUWER, regulation 5). A CMMS full of half-closed jobs can count against you.
Inspection records are a firmer duty. Where PUWER needs an inspection, the result must be recorded and kept until the next one (regulation 6). The format is up to you. The code says records "can be handwritten or stored electronically". It suggests what to include: the equipment and its ID, its location, the date, who did it, any faults, the action taken, who the faults were reported to, and when the repair was done (L22, paragraph 103). Check any system you look at can hold all of that against an asset. Our post on PUWER inspections goes through it in more detail.
For lifting equipment, electronic records are fine too. The LOLER code of practice says a computer system "should be protected from unauthorised alteration" and must be able to produce a written copy (L113, paragraph 373). So ask the vendor two questions. Can a closed record be changed, and does it show who changed it? Can I print a clean report for an inspector?
What to avoid
Some warning signs come up again and again.
Long setups. If the vendor talks about a six-month rollout before anything goes live, ask what you'll have in month one. Start with one line and grow.
Pricing that punishes use. Per-user pricing sounds fine until you want operators raising fault reports and cover engineers closing jobs. Ask what it costs to have everyone on it.
Mandatory fields everywhere. Every required box is a reason not to log. Keep the must-haves to what you'll actually use.
No way out. Ask how you get your data out, in what format, and what it costs. Your breakdown history is yours. Make sure you can take it with you.
Features you won't use. Predictive analytics and IoT dashboards look good in a sales meeting. If you don't yet have clean breakdown data, they've got nothing to work with.
The system is the easy part
The hardest part comes after you sign. The system only works if people use it, and people only use it if they see something back. If a fitter logs a repeat fault and nothing changes, they stop logging.
So close the loop. Use the data in the weekly meeting. Show the team the repeat failures they logged and what you did about them. Our post on getting engineers to use the CMMS covers that side, and a breakdown log that actually tells you something covers what to capture.
And be honest about time. Loading assets, writing PPM tasks and cleaning old data takes weeks of engineer time. If your team is already stretched on breakdowns, that work slips first. Plan for extra hands while the setup's being done, or it never gets finished.
What to do on Monday
- Write your must-have list on one page before you speak to any vendor.
- Pull the breakdown history for your worst machine and see how much of it is usable.
- Ask every vendor for a one-line trial, and give the phone to your night shift.
- Check who will log jobs on the shifts covered by contract engineers.
* Paragraph numbers refer to L22, fourth edition (2014, amended 2018), and L113, second edition (2014, amended 2018), correct as of 26 October 2026. Check PUWER on legislation.gov.uk and HSE's maintenance guidance for the latest.