I’ve always been fascinated by systems, almost to a fault. When I enter an environment, I can’t help but try to understand how it works—what’s flowing where, what’s constraining what, what’s just assumed to be “the way it is.” Some people count ceiling tiles. I map systems.
I was recently in a retail store that essentially takes “stuff” from other places and resells it. You’ve been in this store. Inventory everywhere. Useful things, even—if you can find it. There is almost zero chance anyone working there knows exactly what is in the store at any given time, where it is, or how much of it exists. They just know it’s somewhere—and when it looks like there’s less, they bring in other stuff.
Now layer on a second constraint. Imagine the checkout system only allows you to sell certain things at certain times. A restaurant where chicken is only available at 5pm, steak at 6pm—not because that’s when customers want it, but because that’s how the system is configured. The chefs have preferences, and they spent an enormous amount on that checkout system and need to make it work.
It “works.” Except you’re not hitting your goals, and the instinct is to ask for more capacity—more chefs, more inventory, more hours, maybe another template.
Healthcare runs on this same logic, at scale.
The hiring math doesn’t work. The population needing care is growing faster than we can train providers to serve it. That’s not a recruiting problem — it’s a structural one. Which means optimizing how we deploy the workforce we have isn’t a nice-to-have; it’s the only lever we can actually pull in the near term. Several reports in the last year, including from the AAMC, project a provider shortage of upwards of 86,000 in the next decade. Even recovering 2-3 points of that gap through better utilization would represent tens of thousands of additional patient encounters
Black boxes are only good for airplanes.
A useful frame for data’s impact on long-term provider capacity is the alignment of capacity (supply), demand, and value. Provider time is finite. Demand varies by acuity, type, and source. Value determines how they should interact. A decision in any one area affects the other two—every decision affects capacity. “Good” gets measured at the system level, not the department level. Toyota’s production system is the example most of us have read about: sales, production, and finance aligned around shared incentives and what good looks like.
An aligned team needs an operating system to collaborate and work from. Dashboards with historical stats, or static point-in-time views from a working team, come up short in turning data into an asset. Without this, it becomes hard to negotiate reality. It’s not that everyone’s perspective isn’t true—it’s just uncoordinated truth, like navigating New York City with no stop signs or traffic lights.
And visibility alone isn’t enough. If you have a thermostat in your house that tells you the temperature inside and what the weather is going to do but doesn’t let you turn on the air conditioner or heater, you have visibility with no control—and the temperature keeps dropping. Data, even a sophisticated model, doesn’t elevate a supply/demand strategy on its own. You need the ability to shape demand, take or automate action, and have the system learn in a way that gives you time back—capacity, ideally. It’s a journey from reacting to whatever shows up to proactively iterating on your system, but that’s where the real progress comes.
Provider experience is the lever. How a clinician’s time is shaped matters—to them, and to the effectiveness of the system. Understanding what truly engages them and then using technology to enable that is the control point for keeping capacity, demand, and value in balance. The right technology doesn’t dictate – it enables. That’s the distinction that makes or breaks this.
Not easy, but possible. An approach to consider:
- Align leadership incentives across capacity, demand, and value. A cross-functional team—Ops, Marketing, Referral, Network Integrity, Business Development, Rev Cycle, Finance, IT—working toward the same goal of optimizing the system.
- Enable that team with a real-time operating system. See it, scenario-plan against it, choose the highest-probability play, automate actions, stay outcomes-centric.
- Shape demand by connecting access points to the operating system. Align capacity in real time to each channel—website scheduling, call center, referral—based on acuity. Automate backfilling of cancellations. Use AI to navigate to top of license and respect provider preferences.
- Make learning the default mode. Use AI to surface insights—where templates are out of balance, where demand shape needs work, staffing optimization—and build iteration into the leadership team’s rhythm.
- Engage clinicians as co-designers of the capacity engine – not just end users of it. Understand what they actually want, then embed it into the intelligence system.
When a fully optimized system runs out of effective capacity, we can be confident in how short we are—and what specific hiring will do to maintain access within each community.
It’s possible to eat steak and chicken at 5pm, have the chefs be happy and serve more meals, and have the business make money and feed people well. “Ands” are almost always better than “Ors.” We can work toward a better operating model to reduce the gap —and we can plan for hiring solutions to fill what remains.
