WHY HASKELL STOPPED THROWING PMs INTO THE DEEP END
Ask anyone in construction what a project manager (PM) is responsible for, and you will get a list that never seems to end. Budgets and buyout. Schedules and submittals. Contracts, client relationships, subcontractor performance, risk, safety, billing…the list goes on. The PM is expected to know all of it and do all of it well. It is one of the broadest, hardest roles on the org chart, and the volume of what a person has to master is staggering.
So how has the industry historically trained people for it? For the most part, it hasn’t. The traditional method is to throw someone into the deep end and see whether they swim. You hand a capable assistant project manager more and more responsibility, wish them luck, and let live projects do the teaching. Some people make it. Plenty struggle longer than they should, learning hard lessons on real jobs with real money at stake. It was called a rite of passage. It was really a training strategy of doing almost nothing and hoping for the best.
That approach is getting more dangerous every year. Projects keep getting more complex, and workforce shortages mean you can’t simply hire your way around a gap. The person you already have needs to be able to step into whatever the project demands sooner and more reliably than the sink-or-swim model ever delivered. You cannot afford to let good people drown on the job while they figure it out.
The Strategy: Rotation as a System
Our model cycles assistant project managers through the core functional disciplines of a construction project: estimating, scheduling, financial management, document control, field leadership, procurement, safety, and quality. In this time, they are really doing the job, with guided supervision. The premise is straightforward. A project leader who has sat in each seat can step into any phase of a project with confidence, because they earned that confidence through practice.
The Framework: 4 Ingredients in Every Rotation
The part worth adopting is how a single rotation is built. Early on, we learned that “go shadow the estimating team for a few weeks” produces wildly different results depending on who is doing the shadowing and who is hosting them. So we standardized the format. Every rotation now combines four components, and a rotation isn’t complete until all four are satisfied.
1. Self-guided fundamentals. Before a participant rotates into a discipline, they work through self-paced foundational content that gets everyone to a common baseline. This protects the time of the experts hosting them.
2. Virtual instructor-led sessions. If fundamentals cover what something is, the instructor-led sessions—run virtually so we can pull participants from any office or job site—are for the nuance: why PMs make the calls they make, where the judgment sits, what the textbook leaves out.
3. Observation task lists. Each rotation carries a specific checklist of things the participant must observe and document in the live environment, tied to whatever real project they are embedded in. The task list converts “I spent time with the team” into “I observed and recorded these 11 discipline-specific events.”
4. Required field observation hours. Knowledge does not transfer from a screen alone. We set a minimum number of hours each participant spends in the field, observing the discipline in its actual habitat, before the rotation counts as finished.
Any one of these on its own does not create an ideal learning environment. A self-guided course with no field time is just eLearning. Field hours with no structure deliver wildly different outcomes between learners. Stacked together, with the task list as the spine, they produce something repeatable. Two participants who rotate through any rotation six months apart now have a comparable experience, and we can say what “completing a rotation” practically entails.
The Challenge Nobody Warns You About: Proving It
I will be honest about the early version of this program: We believed in it before we could prove it. The rotations developed strong PMs, leadership backed the effort, and the results looked right. None of that is the same as evidence. A program defended by conviction alone lasts exactly as long as the people doing the convincing.
So we are refreshing the program with measurement in mind. We are constructing a formal framework that moves past satisfaction scores and completion rates toward the questions that matter. Are participants applying these skills in their day-to-day jobs? Is that application showing up in business outcomes such as project performance, retention, and our ability to deploy people where they are needed?
The hardest thing to measure is also the thing the program exists to build: judgment. A multiple-choice quiz can tell you whether someone memorized the order of precedence in a contract. It cannot tell you whether they would make the right call on a project when a fire drill pops up at 4 p.m. on a Friday. To combat this, we are testing artificial intelligence-powered assessments designed to probe critical thinking and decision-making rather than recall. The participant works through a realistic scenario and makes a judgment call. The assessment pushes back, asking why, introducing a complication, and surfacing how the reasoning holds up under pressure. Done across the arc of the program, the goal is to show growth in how someone thinks, not just what they know.
To be clear: This is a test at this stage, nothing more. We are piloting it, watching how it behaves, and treating the early output as something to learn from. If it works, it gives us a way to demonstrate higher-order growth that traditional assessments have never captured well. If it doesn’t, we will have learned that early and cheaply.
The Bigger Bet
You cannot specialize your way out of a talent shortage. The construction industry needs people who can function across the entire lifecycle of a project, and you don’t get those people by accident. You build them on purpose, through structure. A rotational model produces well-rounded leaders. A serious measurement framework proves it. We are committed to both, and we are learning a great deal in the building of it.
That, more than any single result, is the best practice: We stopped guessing, and we are working to improve.