Most Operations Can't Tell You How Many People They Actually Need

The technician shortage is real. The numbers this year are ugly: industry research has shops running about a fifth of their technician roles unfilled, more than 60% of new techs arriving with no formal training, and thousands of dollars and hundreds of hours required to get each one productive. Fleets are competing for the same people as construction, utilities, and manufacturing, and everyone's paying more to lose the same race.

All of that is true. And it's still hiding a more expensive problem, because most multi-site operations can't actually tell you how many people they need in the first place.

That's the part nobody puts in the trend pieces. Before you can lose the hiring race, you have to know what you're hiring for, and a surprising number of operations are running on a number that's really just a feeling. Ask a regional leader why a given site has nine technicians instead of seven and the honest answer is usually some version of "that's what we've always had" or "the manager there said he was drowning." That's not a staffing model. That's a guess wearing a budget.

The shortage is real, but it's covering for a counting problem

When the labor market is loose, gut-feel staffing survives, because the cost of being wrong is cheap and fixable. You over-hired? Attrition quietly corrects it. You under-hired? You post a req and fill it in three weeks. The slack in the market hides the fact that you never had a method.

A tight market removes the slack. Now every seat you misjudge is expensive in both directions, and the misjudgment compounds because the correction takes longer. The shortage doesn't create the staffing problem. It just stops paying to cover it up.

So the operators who are panicking about the shortage and the operators who are quietly fine are usually not separated by how well they recruit. They're separated by whether they can answer a simple question with a number instead of a shrug: how many people does this specific site need to hit its volume at the margin we committed to?

Gut-feel staffing fails in both directions at the same time

Here's the part that surprises people who haven't run a multi-site operation. When you staff on anecdote, you don't just get understaffing. You get understaffing and overstaffing simultaneously, across the same network, in the same week.

I watched this play out across a nationwide last-mile delivery fleet. Field managers were sizing their own teams off instinct and recent pain. The managers who were vocal and good at making the case got more headcount. The quieter sites got starved. The result wasn't a network that was uniformly short. It was a network where some locations were bleeding revenue because they couldn't cover the work, while others were eroding margin because they were carrying people the volume didn't justify. Both problems were invisible at the top, because they averaged out to a headcount number that looked roughly right.

That's the trap. The aggregate looks fine. The distribution is a mess. And you can't recruit your way out of a distribution problem, because you don't have a shortage of people, you have a shortage of knowing where they go.

A staffing model is mostly just a ratio and a refresh rate

The fix isn't complicated, which is why it's frustrating that so few operations do it. A workable staffing model needs two things: a ratio that ties headcount to the actual driver of work, and a refresh rate that keeps the ratio honest as conditions change.

For that delivery fleet, the driver was vehicles. So the model was built on vehicle-to-technician ratios, and the ratios weren't uniform, because the work isn't uniform. Light-duty cargo vans ran around 110 vehicles to one technician. Heavy-duty box trucks, which break differently and take longer to service, ran closer to 55 to one. You apply the right ratio to the actual fleet at each site and you get a defensible headcount target instead of a feeling. Then you hire against a billability target a notch below 100%, so you've got ramp room and you're not betting the site's margin on perfect utilization from day one.

The ratio is the easy half. The discipline most operations skip is validating it site by site before rolling it out. A model handed down as a mandate gets resisted and gamed. The same model walked through with each location's manager, so they can see the math and push back on it with their own reality, gets owned. That's the difference between a spreadsheet nobody trusts and an operating standard the field actually defends. It took months of validation before a single hire moved, and that was the part that made it stick.

The ratio you built last year is already wrong

Now the seeing-around-corners part, because this is where a static model quietly betrays you.

That same delivery network is electrifying fast. Amazon now has more than 50, 000 battery-electric delivery vans on the road, over 30, 000 of them Rivian EDVs, and they're roughly halfway to a stated goal of 100, 000 by 2030, with new longer-range and all-wheel-drive variants rolling out this year. If you run maintenance for a fleet like that, your vehicle-to-technician ratio is not a fixed constant. It's drifting under you in real time.

An electric delivery van doesn't generate the same maintenance work as a gas or diesel one. Fewer fluid services, fewer of the routine mechanical jobs that filled a technician's day, but a different and scarcer skill set around high-voltage systems, battery health, and diagnostics. The labor isn't gone. It's redistributed, and it requires people you can't pull off the same bench. I led the maintenance strategy for the electric delivery vehicles in that fleet, and the lesson was blunt: the ratio that was correct for a diesel-heavy site is wrong the moment that site's fleet mix shifts, and it shifts faster than your headcount plan refreshes.

So the operator who built a beautiful staffing model two years ago and never touched it is now confidently wrong. The number looks rigorous. It's stale. Which brings me to the failure that taught me to respect the refresh rate more than the ratio.

Stale data is more dangerous than no data

We had a site where the client's projections showed about 100 additional vehicles coming. Headcount tracks volume, so I hired ahead of it, onboarded a technician, got the site ready. Then 20 units showed up. Not 100. The site couldn't carry the extra person at that volume, and we had to let them go. That one's on me, and I owned it.

The root cause wasn't the projection being optimistic. It was that the data feeding the decision refreshed monthly, which meant every hiring call was being made on numbers that were already weeks stale by the time anyone acted on them. The fix wasn't a better forecast. It was getting the client to send deployment data weekly instead of monthly, which pulled the hiring cycle from ten weeks down to six and let us respond to what was actually happening instead of what we'd guessed a month ago.

That's the uncomfortable lesson under all of this. A confident decision on stale data is more dangerous than an admitted guess, because the staleness is invisible. Everyone treats the number as current. Nobody prices in the lag. And in a tight labor market, the cost of acting late on bad data is a person's job and a quarter of margin, not a quick repost.

Monday morning application

If you run an operation: Pick your most labor-intensive function and write down the one variable that actually drives the work, then the ratio of that variable to headcount. If you can't state it as a number, that's your week's project, not the recruiting req. Then ask how old the data feeding that ratio is. If it refreshes monthly and your conditions move weekly, fix the refresh rate before you touch the model.

If you advise operations: Don't accept an aggregate headcount number. Ask to see the distribution by site against a stated ratio. The gap between "we have about the right number total" and "every site is staffed to its actual driver" is where the recoverable margin and the hidden revenue leak both live.

If you're earlier in your career: The next time someone justifies a headcount with "that's what we've always run, " treat it as an open question, not an answer. Find the thing the work actually scales with. Learning to tie people to a driver instead of a habit is one of the most portable operating skills you can build, and almost nobody does it on purpose.

Until next Tuesday,

Mason

---

Mason Gray writes weekly on operations leadership at mid-market companies. He advises a few operating teams (Decion Technologies) and is in conversations about senior operations roles. Reply to start one.