I argued in Are We Screening for the Right Things? that the years filter broke. This piece is about why it broke, and what should replace it.
Every posting for a plant operations manager asks a candidate a question before they ever speak to anyone. How many years do you have?
Most industrial postings still do not debate that question. It got copied from the last posting, which got copied from the one before it.
For a long time, it was a reasonable question, because it stood in for better ones. Nobody hires twenty years. They hire what twenty years is supposed to guarantee. Some of that is knowledge: where the answers are, how the equipment behaves, what was tried before and why it failed. The rest is judgment: having made enough hard calls that the next one is not new.
Asking about years used to be a fair way to ask about both. That is no longer true.
Start with knowledge. Procedures, specs, maintenance histories, and the troubleshooting steps that used to live in one person's head can now be written down, and AI can make them searchable in seconds for whoever comes next. Knowing where the answer is used to take a decade. It increasingly does not.
What does not transfer is the judgment. Knowing which answer applies right now, and when the system is wrong. The call to shut a line down on a hunch. The read on a crew that is about to walk. That part is still built one hard decision at a time. Asking about years cannot tell the two apart.
Careers had steps for a reason. Technician, then lead, then supervisor, then manager. Each step handed someone a bigger problem and a higher cost for getting it wrong, and the years were how long the climb took. When we asked how many years, we were really asking how many steps.
That link is weakening from both ends. Some people stay on one step for a long time and still collect the years. And the bottom steps are changing. Much of the routine work that taught someone how a plant runs, before anyone trusted them to run part of it, is the same work AI and automation are absorbing. Stanford researchers tracking payroll data found that since generative AI spread, employment for early career workers in the most exposed occupations fell by double digits, while experienced workers in the same jobs held steady. The clearest evidence so far comes from office and technical roles. Fewer bottom steps means fewer people building judgment the slow way. That makes the people who already built it worth more, not less.
Now add the labor market. The US labor force has contracted by roughly 700,000 people this year. In technical and industrial hiring, the qualified pool for a given role was already thin. New manufacturing capacity going up in regions that have never supported it makes it thinner. Every question on a posting now removes people a company cannot afford to lose, and the years question is usually the first one asked and the last one dropped, often months later, after the search has already failed.
Here is what that question misses. Two candidates, each with twenty years in operations. One took plants through capacity ramps and rebuilt a maintenance program after the last one fell apart. The other ran one stable line, very well, for two decades. Neither is better as a person. They are built for different seats, and they give the same answer.
A third candidate, a Staff NCO with twelve years in, owned an aviation maintenance program through a deployment. A missed standard had real consequences there, and the plan broke more often than it held. The military still hands out responsibility one step at a time, often earlier than a civilian career would. He climbed the same ladder, just faster. Count the hard decisions inside those twelve years and he may have more than either of the first two.
His answer screens him out before anyone hears the rest of it.
None of this means experience has stopped mattering. The supervisors retiring out of industrial companies right now are leaving with judgment that took decades to build, and that loss is a large part of why this market is so tight. Experience is not the problem.
The question we use to measure it is.
So what are the right questions? They start with the seat, not the candidate. What does this role need to accomplish in its first year? Is it a ramp, a turnaround, or a steady operation that needs to stay steady?
Then the questions for candidates follow from that answer. Which steps have you climbed? What have you run, and under what pressure? How many times did the plan break, and what did you do next?
And two questions years never answered at all. Can you explain clearly what happened and why? Will you put yourself on the line for a call before you are sure it is right?
Those matter more now, not less. When the answer can be looked up, the person worth hiring is the one who can make a crew understand it, and stand behind the decision when it counts.
Years used to answer those questions for us. Now we have to ask them.