Key Takeaways
- The deepest reason to leave paper is memory. Efficiency is what a system gives you this quarter. Keeping what your most experienced people know is what it gives you for the next twenty years.
- A departing expert can't hand over what they know on request. Expertise gets retrieved by situation: standing at the machine they remember, sitting at a blank page they don't.
- Capture work when it rides on work already happening, which means it starts years before anyone hands in notice.
That's why exit interviews and handover spreadsheets produce so little. They ask for a download from a person whose expertise doesn't live in that form. The real value of a system shows up years later, when a veteran retires and everything they learned about every machine stays behind.
What actually walks out when an experienced technician leaves?
Four things leave with them, and none is on the org chart. What goes is the accumulated read of a plant: which machines lie about their condition, what the last real fix was, who to call when a part is discontinued. A competent replacement arrives with none of it.
- The quirks. What a machine does in the week before it fails. The noise, the drift, the way it runs fine until the product line changes.
- The failure history. What went wrong, how often, and which repair held versus which one bought three months.
- The workaround. The sequence that gets a stubborn asset back up, invented once at 2am and never written anywhere.
- The supply knowledge. Which supplier ships fast, which part number superseded which, what fits when the original is gone.
Why doesn't asking them to write it all down work?
Because a blank page asks for recall and the job asks for recognition. Hand someone a form and they produce a thin, dutiful list of the obvious. Put them in front of a pump they've rebuilt four times and the specifics come out in order, because the machine is doing the remembering with them.
Capture fails the same way whenever it's built as a separate task. If a workflow asks a technician to do something they'd never do with a clipboard, they stop doing it, and every system we've watched die has died at that seam. The write-up that waits until after the shift is the write-up that stops happening in week three.
A maintenance director with thirty years on the floor put the resistance in the right place for us. Mechanics want to turn a wrench, and what they're refusing is friction. When logging a job takes one tap instead of nine fields, the same people use the system. Worth noting: the person whose memory is worth most is usually the one least interested in a new way of working.
What does maintenance knowledge management actually look like day to day?
It looks like three ordinary places filling up while the work gets done. Nobody sits down to write a manual. The technician closes a work order, ticks a checklist, adds a photo, and the operation accumulates the record it was missing.
The asset logbook pulls key information from work orders onto one screen, so a machine's history reads as a single story. At Tricel, technicians review that history to understand an asset's condition before starting a repair. The next person opens the machine already knowing what the last one found.
Checklists are where one person's judgement becomes everyone's method. Holistic Industries found one facility with significantly better water treatment than the rest. That facility's cleaning process became a checklist on the maintenance schedule at every site, and all sites now have excellent water treatment. One person's standard, applied everywhere, without that person in the room.
Their director of project management put the result plainly: "Now that we have the CMMS, all that information resides in the system and stays within the business. It makes Zoidii the employee with all the knowledge that will never leave."
What can a system never capture?
Judgement. A record tells the next technician what happened, what was tried, and what held. It stays silent on how the expert knew to look at the gearbox first. Diagnosis is pattern recognition built over years, and no field on a form holds it. Say that out loud rather than pretending otherwise.
What the record does is shorten the apprenticeship. Ten years of failure history lets a newer technician reason from what this specific machine has done, rather than from what machines like it usually do. The instinct still has to be built. It gets built on evidence instead of on three years of being surprised.
Where do you start if someone is already leaving?
Start with the assets only that person touches. The list of equipment where the operation depends on one person's memory is usually short, and everyone already knows what's on it.
Then work three passes with them.
- Fix the asset records. Create or correct them for those machines, so specifications and manuals sit somewhere other than a drawer.
- Turn repeat failures into checklists. In their words, capturing the sequence they follow rather than the one the manual describes.
- Log every remaining job. Every work order they run between now and their last day, because those are the last entries anyone gets.
What survives is what got written down while the work was being done. Everything else is a conversation you meant to have.
Ask any vendor to show you where a technician's note about one specific machine ends up, and who sees it the next time that machine gets opened. If the answer is a search box and a hopeful expression, the system will hold your records without holding your memory. The day someone hands in notice is the day this method stops being available. Maintenance knowledge management is what you were doing for the two years before that, or it's a spreadsheet somebody fills in on their last Friday.
