Optimize One Part, and the Whole System Pays For It

A chain is no stronger than its weakest link

Eliyahu Goldratt, Author of The Goal

One of the more surprising lessons I've learned in leadership is that a decision can be completely correct for one part of an organization and still be wrong for the organization as a whole. That sounds obvious in theory. In practice, it's remarkably difficult to spot, because it doesn't announce itself — it shows up quietly, weeks or months later, in numbers nobody was watching at the time.

Most organizations aren't single systems. They're networks of interconnected ones, each with its own objectives and constraints, rarely operating independently. A hospital makes this especially visible: a patient moving from outpatient consultation to admission, surgery, intensive care, rehabilitation, and discharge passes through dozens of teams and systems, and when one becomes constrained, the effects ripple across everything else. Nothing has taught me how easily one department's solution becomes another department's problem more than trying to expand ICU capacity.

The constraint is rarely what people think it is

A few months after opening, we found ourselves discussing the same issue repeatedly: our inability to expand ICU services. On paper, we had over 30 ICU and High Care beds available, with the physical space and most of the equipment in place. We could reliably operate five. The question was simple — what was preventing us from using the rest?

Like most organizations, we initially chased the problem people talked about most. A recurring complaint was the shortage of infusion and syringe pumps, and it came up often enough in meetings and WhatsApp threads that it hardened into accepted wisdom: ICU capacity was constrained by equipment. We had never actually measured it. Once we tracked the data properly, a different picture emerged — equipment shortages happened occasionally, but they weren't the bottleneck. The real constraint was nursing. We had roughly 11 to 12 qualified ICU nurses covering 12-hour shifts, and maintaining safe staffing ratios meant those nurses could support only a handful of beds, no matter how much physical capacity sat unused down the hall. We had mistaken the most discussed problem for the most important one — the weakest link in the chain was never where the conversation kept landing.

The fix wasn't more pumps. It was building our own ICU training pipeline to grow the number of nurses qualified for critical care — a solution that takes months rather than days, but addresses the constraint that was actually holding us back. The issue people talk about most is not always the issue holding the system back.

Improving one metric can worsen another

The nursing shortage led to a second lesson. Many of our nurses commute significant distances — some up to two hours each way — and under an 8-hour shift structure, meeting their required hours often meant working six days a week. Several proposed moving to 12-hour shifts instead: same total hours, fewer working days, more rest, less commuting. The logic held up on paper. At 100 nurses, six 8-hour days gives you 600 man-days a week (equivalent to 4,800 hours); four 12-hour days gives you 400 (also equivalent to 4,800 hours) — a real reduction in scheduled burden, achieved without cutting anyone's pay or hours. The total labor hours remained identical, but the number of available shifts and opportunities for overtime changed materially. We agreed to the change.

What we hadn't anticipated was overtime. It's far easier to ask someone to extend an 8-hour day than to stay for a second consecutive 12-hour one, and once we made the switch, willingness to pick up an additional shift dropped sharply. The scheduled man-days on paper stayed exactly as planned; the practical flexibility of the workforce did not. Staff satisfaction went up. The hospital's capacity to safely absorb more patients went down. Neither outcome was visible when we made the decision — both became obvious only afterward. The lesson wasn't that 12-hour shifts were the wrong call. It was that every local optimization carries the risk of quietly creating a new constraint somewhere else in the system.

Every fix creates a new problem

The challenge now isn't identifying the issue — it's deciding how to respond. The move to 12-hour shifts genuinely improved many nurses' lives, and reversing it outright would likely be read as leadership taking back a benefit rather than solving an operational problem. That perception matters as much as the underlying math.

So the path forward isn't a simple reversal. We're building the case with data first — tracking productivity and overtime availability before and after — so that any correction is grounded in evidence rather than management preference. The eventual answer may be a hybrid model, or targeted 8-hour shifts in specific roles and departments, rather than a wholesale policy change; hiring enough additional nurses to remove the tradeoff entirely isn't realistic without significant added cost. We don't yet know exactly how this resolves, and I think that's worth stating plainly rather than dressing it up as solved. Leadership often means operating in the space between recognizing a problem and fully understanding its solution — and managing the fact that the fix for one constraint frequently introduces another.

Running the whole system, not just its parts

The same pattern shows up everywhere in a hospital, just less visibly — in drug procurement, training pipelines, benefits administration, digital systems, the cadence of communication that keeps departments aligned. Each function can optimize itself while unintentionally making life harder for the organization around it.

Running a hospital, in that sense, has more in common with running an airport than with managing a set of departments. Air traffic control coordinates hundreds of flight schedules a day while baggage handling, security, fueling, and retail all run in parallel — each its own system, each capable of running well in isolation and still degrading the whole if no one is watching how they interact. A hospital mirrors that structure closely: outpatients moving through clinics, admissions arriving through emergency or scheduled surgery, operating theatres running procedures, ICU providing critical care, and behind all of it a building being maintained, drugs being depleted and replenished, and people being managed and kept motivated enough to do it again tomorrow.

Optimizing one department is relatively easy. Optimizing the institution is much harder — it requires continually asking whether the institution as a whole is getting stronger, not just whether the one part in front of you is performing better. Organizations don't succeed when individual departments win. They succeed when the entire system does.