Capturing the knowledge that runs your system before it retires

You already know the operator. Every small system has one: the person who can coax Valve 7 open in a hard freeze when nobody else can, who knows which meter reads low and why, who walks into the pump house at 2 AM and puts a hand right on the fitting that always fails first.

None of that knowledge lives on a spreadsheet. It lives in one person's head, and it walks out the door with them the day they retire.

Running the system: rung 2 again.

Small systems don't just lose a person when a longtime operator leaves. They lose the map that never got drawn, the workaround that only worked because of a trick nobody wrote down, and the reason a certain valve gets left cracked a quarter turn. Rebuilding that from memory, if you can rebuild it at all, costs more than any raise that might have kept them.

What retirement actually costs a small system

This isn't a slow drain. Utilities are set to lose 30 to 50 percent of their workforce to retirement within the next ten years. Half of water-utility workers are already over 44, the average retirement age is 56, and the people retiring have averaged 24 years at the same utility.

That last number is the one to sit with. Twenty-four years is long enough to build a mental map of every quirky valve, every buried line nobody surveyed, and every fix that only works one particular way. None of it transfers on a two-week notice.

Knowledge isn't one thing, and writing it down isn't enough

There are four kinds of knowledge, and only one of them is easy to keep. Explicit knowledge is already written somewhere, a valve card or a manual. Tacit knowledge is know-how, the exact feel of how hard to lean on a stuck gate valve. Implicit knowledge is known but never shared out loud. Evident knowledge is something everyone can see a person doing, but the person doing it never thinks to explain why.

A binder full of written procedures sounds like the fix. It isn't, by itself. Think about a bicycle: you can hand someone a perfect written description of how to ride one, and they still can't ride it.

Some knowledge only transfers by doing it alongside someone, watching, and asking questions. That is why interviewing the person who has the knowledge matters as much as any manual.

The minimum viable record

You do not need a perfect archive. You need five things captured for every asset and every event:

• when and where it happened, marked on a map
• what asset was involved, with an ID
• what happened and why, or "unknown" if you genuinely don't know
• what was done about it, and any follow-up
• a photo

Unknown is allowed. Guessing is not. Asset ID is the glue that holds a record together, and a photo beats a paragraph every time.

For a main break specifically, capture the Five W's and how:

• who worked it
• what the issue was
• when it happened
• where, precisely enough to find the segment again later
• why it likely happened
• how it got fixed, including materials and how long the water was off

Keep each record to two minutes or less to fill out. Use a dropdown instead of free text for anything that repeats, like material, status, or cause. Ten different spellings of "ductile iron" turn a database into a junk drawer nobody trusts, and a database nobody trusts can't back up a funding request either.

How to interview the operator who's leaving

Don't wait for the retirement announcement. Run a version of this four-question check once a year, with every operator, not just the ones near retirement:

• who has knowledge that matters
• what knowledge is actually at risk of walking out
• what happens to the system if it's lost
• what you're doing about it now

Frame it as helping someone pass on their legacy, not as a performance review, and use it only for planning, never to make a decision about a specific person's job.

When someone is actually leaving, don't try to capture everything they know. A full knowledge audit typically finds that only 20 to 25 percent of what a longtime operator carries is worth writing down; the rest is repetition or things any trained operator already knows. Ask what breaks that nobody else can fix, what shortcuts exist that aren't in any manual, and what would confuse a stranger standing exactly where they're standing.

One utility videotaped a retiring operator walking through the repair steps for a rarely-touched part of the system that was brutal to diagnose blind. That tape outlasted the operator's own tenure.

If a retiree comes back as a consultant or part-time hire, make the knowledge transfer a condition of the rehire, spelled out up front, and pair them with a mentee outside their old chain of command so it doesn't quietly turn into them running things again.

Make it part of the job, not a someday project

A document-everything initiative dies the moment the next emergency call comes in. What survives is documentation built into the work operators already do, filled out as part of closing the ticket, not tackled later as a separate project nobody has time for.

The barrier is almost never willingness. Operators don't hoard knowledge, they hoard their time, because writing something down competes with everything else on the truck that day.

Keep the record about two clicks away, or people will build their own workaround around it. And lean on the loop that keeps it going on its own: when good records sit at a crew's fingertips in the field, their own next call gets easier, so they keep updating them, and coworkers hold each other to it because everyone is relying on the same data for the next break.

What it costs to skip this

The dollar case is not abstract. Replacing a certified treatment operator costs a minimum of about one and a half times their salary. A separate study of information workers put the cost of not being able to quickly find information at roughly $5,300 per employee per year, and that number only grows once the one person who knew where things were is gone for good.

Real systems have paid this price. One utility lost its maps and customer records in a wildfire and never recreated them. Another lost its sole operator with no passwords and no accounting records left behind.

And it isn't only fires and sudden exits. Operators swapping stories about the oldest thing at their plant keep landing on the same detail: handwritten logs from the 1990s in bound composition notebooks, half of it in cursive, naming operators and supervisors nobody currently on staff had ever met. Those notebooks were the only real record of who ran the plant in that era. When the people who wrote them retire and the notebooks get tossed, the memory goes with them.

There's a myth worth naming directly. Some operators worry that documenting what they know makes them replaceable.

It's the opposite: undocumented systems create burnout, nonstop after-hours callouts, and turnover that then hands the next person a knowledge cliff of their own. Job security comes from running a reliable system, not from being the only one who knows how it works.

Get the handoff in writing, before you need it

If you take over a system from another operator, a contractor, or a private owner, do not assume you will get good records. In practice, incoming crews often receive nothing more than an outdated paper map and a meter-read sheet. Put the requirement in the contract itself: name the specific records that must be kept, and if they're still on paper, require them to come in with each invoice instead of showing up as a surprise on the last day.

And when a gap turns up later, a year of samples that never made it to the lab, a valve nobody can find on the map, report it up the chain immediately and in writing. An operator who finds a mess and reports it right away is in a very different position than one who quietly patches around it.

Next on the ladder: Staffing and turnover, it's a budget decision