Every plant has one.
The guy who has been there thirty four years and can tell you the bearing is going out by the sound it makes through a wall. The woman who knows which customer will accept a part half a thousandth over and which one will send the whole lot back. The setup nobody else can run, on the machine nobody makes anymore.
Ask what they know and they will tell you they do not really know anything special. They have just been doing it a long time.
He retires in April.
A written procedure captures the steps. It cannot capture the thing your best operator hears before the machine fails. How to film that before he retires.
Why the written procedure did not work
Most companies have tried. Somebody spent a month writing procedures, they went in a binder, and nobody has opened it since.
That is not because the writing was bad. It is because of what a written procedure can and cannot hold.
A procedure captures the steps. The steps are the easy part. Those are mostly in the manual already.
What is missing is the judgment. When to stop. What it should sound like. Which of the four things that could be wrong is actually wrong, based on something the person cannot quite name.
Try writing down "back off a quarter turn when it starts to chatter." Now try writing down what chatter sounds like.
You cannot. That knowledge does not live in words. It lives in hands and ears, which is exactly why it has never been written down, and it is also why the person who has it does not think of it as knowledge.
What video does that text cannot
You can show the sound.
That is the entire argument. Film the machine when it is running right, then film it when it is starting to go wrong, and a new operator hears the difference in four seconds. No amount of writing gets there.
Same with hands. Where he actually puts his thumb. How hard he pushes. The half second pause before he commits. A new person watching that picks up things nobody thought to mention, because the person doing it does not know they are doing it.
The other thing that matters is who is talking. A procedure written by an engineer reads like it was written by an engineer. The same information out of the mouth of the guy who has run that machine since 1991 gets believed on the floor in a way a binder never will.
The mistake that kills most of these projects
Asking the person to explain what they know.
Sit a thirty four year machinist down and say tell us everything you know about this machine, and you will get about nine minutes of nothing. Not because he is holding back. Because it is automatic. He genuinely cannot see it anymore, the same way you cannot describe how you drive.
Two things that work instead.
Have him do the job while he talks. The hands unlock the mouth. Narration during the actual work produces ten times what an interview at a table does.
Better, have him teach somebody. Put a newer operator next to him and film the two of them. He will stop reciting and start correcting, and correcting is where all the real knowledge is. Watch her hand, no, not like that, hold it like this. That is the footage you came for, and you cannot get it any other way.
This is also why the people who protest hardest usually end up being the best on camera. Nobody freezes explaining their own work. They freeze being asked to perform.
What to film, in order
You cannot film everything, so do not try. Triage it.
Start with what is closest to walking out the door. Not your best people. The ones nearest retirement whose jobs are hardest to fill. Make a list, put a date next to each name, and sort it. That list is your shoot order and it will be uncomfortable to look at.
Then, for each one:
Setups and changeovers only one or two people can do. This is almost always the highest value footage in the building.
Troubleshooting. When it does this, here is what is actually wrong, and here is the thing everybody checks first that is never the problem.
The workarounds. Every plant runs on undocumented fixes. The shim. The order you have to do it in. The thing you do because the machine has been out of true since 2009.
Judgment calls. Which material for which job. When to stop and ask. What a good part feels like.
The equipment nobody supports anymore. If the manufacturer is gone, the only documentation left is in somebody's head.
What this does not do
It does not replace training somebody with their hands. Nobody learns to weld from a video, and anyone who tells you otherwise is selling something.
What it does is cut the time to competent, and make sure that when the person who knew is gone, what he knew is still in the building.
It is also close to the only version of this that people actually use. A binder gets shelved. A two minute clip a supervisor can pull up on a phone standing at the machine gets watched.
The part that shows up on the balance sheet
One more reason, and it is the one nobody thinks about until they are in it.
A company where critical knowledge sits in three heads is worth less than one where it is documented. That is not a marketing claim, it is how buyers and insurers look at key person risk, and it is one of the few parts of it you can fix in advance. There is more on that in what gets lost when a company is sold.
If an owner is thinking about an exit in the next five years, this is not a training project. It is preparation.
The short version
The steps are already written down somewhere. What is not written down is what your best operator hears, sees, and feels, and that is the part leaving in April.
Film him doing the work, or better, film him teaching somebody. Start with whoever is closest to retiring and hardest to replace.
If there is somebody in your building whose knowledge is not anywhere but in his head, happy to talk through what it would take to get it out. That conversation is worth having before you have a date on the calendar, not after.
Dylan Smith is the owner of With A Twist Media in Greensboro, NC. He has been making documentary style video for manufacturers and construction companies across the Triad since 2019.