Your Best People Are Doing Two Jobs

It’s three in the morning. A 68-year-old man, admitted yesterday and stable at midnight, goes into cardiac arrest. His nurse hits the code blue button — the call that brings a highly specialized team in when a patient’s heart stops.

Over a quarter of a million cardiac arrests happen in US hospitals every year. About one in four of those patients leaves the hospital alive. Over 200,000 don’t.

Whether this patient survives comes down to two things. One is the live work in the room: the code blue team running a protocol drilled hundreds of times, hands on compressions, voices calling orders, adapting to every complication.

The other is everything that gave them their best chance: the work done in the hours before. The charge nurse flagged his drifting oxygen in handoff. The pharmacist double-checked the new medication. The supply tech restocked the crash cart at shift change. Each role is a link in the chain. The chain breaks at the handoffs.

This patient survived. Most don’t. For many of those 200,000 deaths, something upstream broke first.

The second job: the compensating work (workarounds, rework, manual fixes) that experienced operators do because the system can’t deliver what it promises. None of it is on their job description or measured, and the system never learns from any of it.

What Anita Tucker saw

In the early 2000s, Anita Tucker, then an operations researcher at The Wharton School, spent months observing hospital nurses across hundreds of shifts. She wasn’t looking for clinical errors. She was looking at the operational system around them. What she found has held up across a generation of research.

Nurses encountered an average of 8.4 work-system failures per 8-hour shift: missing supplies, mislabeled medications, charts not updated, equipment in the wrong place. None of these were the dramatic mistakes that lead to malpractice headlines. They were the steady drip of friction that anyone running a complex operation will recognize immediately.

Then the number. In 93 percent of cases, the nurse solved the problem alone. She grabbed the supply from another drawer, double-checked the medication herself, called the lab for the missing result, found the equipment in another unit. She did not flag it, did not escalate. By the next hour she had moved on to three more failures of the same kind.

Tucker measured the time cost of this work. Forty-two minutes per shift. Nearly an hour of every nursing shift spent compensating for things the system should have done.

In hospitals as in operations, flagging is how the system learns. When a problem gets recorded, root cause analysis becomes possible. Fixes get designed. The next nurse on the next shift doesn’t lose another forty-two minutes to the same broken supply chain. When the problem doesn’t get flagged, the system never knows. The failure repeats. The compensating work compounds.

The flagging trap

The hospital has the forms: improvement systems, near-miss reports, quality reviews. The infrastructure for learning exists. It doesn’t get used.

Four reasons nurses don’t report

They will sound familiar to anyone running operations.

Filing the report pulls the nurse away from the patient in front of her. The system rewards solving the problem in the moment. It does not reward documenting the structural cause.

The form doesn’t ask the right questions. It was designed for incident reporting after a serious adverse event. It was never built for the routine ten-second supply problem that just cost her three minutes. When she tries to fit her real problem into the form’s categories, she gives up.

The fix, if it comes at all, arrives weeks later. By then she has worked around the same problem ninety times. The fix is no longer connected, in her mind, to the original report. Cause and effect are severed.

And no one covers the patients she missed while filling out the form. The compensating work is invisible. The flagging is also invisible. But the cost of flagging is visible immediately — to her, to her team lead, to the patient whose call light went unanswered. So she takes care of it herself, grabbing the supply from another drawer, re-checking the medication, giving the next nurse a heads-up.

People want to help each other.

The workaround becomes the help. And the system never sees any of it.

The second job

This is the second job. The work your best people do because the system can’t deliver what it promises: the compensation, the rework, the workarounds. None of it is in their job description or measured on their performance review. Each piece becomes scar tissue, passed forward, until it’s just how the work runs.

The pattern lives in every operation I’ve looked at, whether it’s customer support, site reliability, revenue ops, or supply chain. The names change but the mechanism is identical. The expert solves the immediate problem. The fix that would have prevented the next one goes unwritten. The system never learns.

Why the system never learns

The cost compounds in three directions.

The person carrying the workaround eventually leaves. When she does, the workaround leaves with her. The next person inherits a system that nominally works and operationally doesn’t. They will rebuild the workaround themselves — slower, with less confidence, more brittle. The institutional capability degrades while the org chart stays the same.

The fix never gets prioritized. The product team, the IT team, the operations team get tickets for fires, not for the slow-burning workarounds. The roadmap reflects what got reported. What got worked-around stays invisible.

And AI deployment hits this wall harder than anything that came before. Every agent or copilot built on top of an operational system inherits that system’s actual behavior, not its documented behavior. The model has no access to the undocumented workarounds — the things your best people know to do that nobody wrote down. The deployment fails. The vendor blames adoption, the team blames the technology, and neither has the right diagnosis. The conditions the AI needed were never named.

The leader’s job

You can’t fix what you can’t see. Making the second job visible is the first move.

Three steps.

See it. Spend a week with your best people doing their actual work, not in a conference room or staring at a dashboard. Watch what they do between the things on their job description. The workarounds will show up within hours.

Count it. How many minutes per shift, per day, per week, are being spent on compensation work? You don’t need a survey. You need someone competent observing for two days. The number will surprise you, and it will tell you which workarounds are the most expensive ones to leave in place.

Decide. Some of the compensating work is the right answer — the system can’t economically be redesigned to handle every edge case. Some of it is technical debt the organization decided to live with. Some of it is fixable in an afternoon and has been sitting there for two years. The decision is yours. The point is to make it a decision, not an inheritance.

Your best people are doing two jobs. The first you’re paying for. The second is the compensation, the rework, the workarounds – yours to make go away.