The Staffing Forecasting Gap: Predict or React
Robert Grand · Battalion Chief who still runs calls
Every Fire & EMS service has that one person. The savant who knows the staffing models. Ask them in July when you’ll hit a crunch and they’ll tell you: August 15, we’ll need three more people. Ask them in November which quarters historically push retirement rates and they’ll map the next two years of hiring. They carry 20 years of staffing patterns in their head. They know when seasonal pressure hits, when academies graduate, when vacation picks cluster, when retirements cluster in July and December. They have built a mental forecast so precise that the battalion chief calls them before proposing any major shift change.
That person is invaluable. And retiring soon.
The Knowledge in Someone’s Head
Shift commanders and battalion chiefs across Fire & EMS already understand the staffing cycles mentally. They know the trends. They know that protected vacation picks mean you can’t prevent August shortfalls just by closing access to a calendar. They know that the shift to 24/72 and D-shift schedules brought retention gains but reshaped when stress hits. They know that the person sitting three desks over has spent 20 years building a staffing forecast that nobody else fully understands.
You can’t replace them while they’re still working. When they retire, the next person has to learn what they carried. Two years of fumbling. Three years of hiring at the wrong times. Five years of building back to the institutional knowledge that just walked out the door. That is not a succession problem. That is a knowledge problem.
What Happens When They Retire
Most Fire & EMS services have never written down the staffing model this person carries. It lives in quarterly reviews, in a spreadsheet nobody else maintains, in a set of decision trees that only live in their head. The patterns are real. The accuracy is real. The method is not documented.
So when they retire, the next chief has to reverse-engineer the patterns from historical hiring decisions. Why did we hire 12 people in March 2024 and only 4 in June? Was it academy timing? Vacation seasonality? Retirement predictions? All three? The answer is in the savant’s head. Without it, the new person defaults to reactive hiring, backfilling when the shortfall appears instead of forecasting when it will arrive.
That hiring lag costs you. It costs you in forced overtime. It costs you in burn-out because you are cycling inexperienced people through stations that should have depth. It costs you in retention because you are running experienced medics ragged during crunch periods instead of having predicted the crunch and hired six months earlier.
The Real Problem Is Not Predicting Sick Calls
Healthcare talks about staffing prediction as if the problem is Monday morning sick calls. That is not your problem. Your service has protected vacation picks. Your people plan their time off. The holiday crowds, the summer peaks, the winter retirements. That part you can see coming. The battalion chief knows it. The shift commander knows it. The savant knows it exactly.
The problem is not forecasting short-term call volume. The problem is documenting the hiring model so that when the savant retires, the next person does not have to spend three years rebuilding what was already known. It is about capturing the decision framework: hiring levels by quarter, by academy output, by retirement patterns, by the interaction between all three.
That is a documentation problem, not a forecasting problem.
The 36-Month Advantage
The Fire & EMS service that codifies its staffing model before the next senior person retires has 36 months of runway. They document the patterns. They build the hiring schedules. They make the forecast teachable to the next operations captain. When that person walks out the door, the new person does not start from scratch. They inherit a model.
The service that waits until after the savant retires will spend three years figuring out what the savant already knew. Three years of sub-optimal hiring. Three years of reactive staffing instead of proactive. Three years of experienced people burning out because you predicted wrong. That is the cost of not writing it down.
Where to Start
You do not need a consultant or a vendor model. You need one battalion chief or operations captain to spend a quarter documenting why hiring decisions were made in previous years. Why did you hire eight in Q2 2023 and five in Q1? What was the academy pipeline? What retirements were forecast? What vacation seasonality showed up? The answers already exist. They just exist in the savant’s head.
Once you have that documented, the next person can learn from it. They can challenge it. They can refine it. But they are building on known patterns instead of starting blind.
The Fire & EMS service that does this first has institutional knowledge after the savant retires. The ones that wait will have three years of fumbling while the next generation figures out what was already known. The difference is not about being early adopters. The difference is about not losing 20 years of staffing intelligence to retirement.
Then AI Can Help
Healthcare did not stop at documenting their staffing patterns. They documented them, then fed that documentation and hiring history to AI systems. The system learned the correlation between academy timing and hiring needs. It learned seasonal variation. It learned the interaction between retirement patterns and recruitment timelines. Now the hospital does not depend on a single forecaster. The forecasting lives in the data and the model. An AI agent learns from the documented patterns and historical hiring decisions, identifying gaps before they happen, surfacing hiring needs months out, and continuously refining the forecast as new data comes in.
This is agentic modeling. The pattern becomes resident in the system. When the person who built the first model retires, the next chief does not inherit a black box. They inherit a documented process that an AI system is actively maintaining and improving based on real data. The forecast stays predictive because the method is no longer in someone’s head. It is in the data.
For Fire & EMS, this means once you document the staffing model, the next step is not hiring another forecaster. It is taking that documented knowledge and building a system that can learn from it, maintain it, and adapt it as your service changes. The tool does not matter yet. The documentation does.
Robert Grand is a Battalion Chief at Eugene Springfield Fire with 24 years of service. He writes Frontline Intelligence, a newsletter on operational doctrine, technology, and leadership in Fire & EMS.
If this landed, share it with someone in Fire & EMS who needs to hear it.
Subscribe
From the floor, not the vendor booth. Two times a week.