What You Reward, You Create

Most organizations don’t pay for prevention. They pay for survival.

There is a kind of work in every organization that produces no story.

It’s the careful redesign of a system so that a whole class of failures becomes structurally impossible. It’s the early conversation that kept a project from going down the wrong path. It’s the architectural decision in week three that prevented six months of pain in year two. It’s the patient, unglamorous, often invisible work of making bad outcomes less likely.

The people who do this work tend to know exactly what they are doing. They also tend to know that no one will quite see it.

I recently read about an engineer at a major tech company who caught a defect that would have triggered an $80 to $90 million product recall. The fix was pretty straightforward. The save was real. Her reward was a “good catch.”

Six months later, the CFO mentioned the avoided cost on an earnings call. No name. No credit. The story moved through the company as a vague piece of good news, a small fact in a quarterly narrative she had quietly made possible.

She received roughly the same recognition an outfielder would have for making a difficult catch. The catch that no one realized had it not been made, the team would’ve lost the game.

I keep thinking about her.

Not because what happened to her was unusual. Let’s face it, it’s all too common. No, it’s because of the comments underneath that story. Engineer after engineer described the same pattern from inside their own organizations. People had stopped reporting pre-incident bugs. They were quietly storing the knowledge — sometimes for weeks, sometimes longer — and waiting for the bug to become an incident. Then they would “discover” it during the firefight, resolve it heroically, and ride a hot Sev-1 ticket into their next promotion.

These aren’t “bad” engineers. These are rational people responding to the actual incentive structure of their organizations. They’re doing what the system asks of them.

They have noticed something most executives haven’t: visible heroism gets rewarded; invisible prevention does not.

This isn’t new.

Scott Adams drew this exact dynamic in 1995. A Dilbert strip from November of that year shows the pointy-haired boss announcing a ten-dollar bonus for every bug an engineer finds and fixes. Dilbert and his colleagues celebrate: “Yahoo! We’re rich!” Wally announces he’s going to write himself a new minivan that afternoon.

The joke wasn’t at all subtle. It wasn’t even new in 1995. And yet thirty years later, most engineering organizations are still doing some version of the same thing.

That tells you something important.

This isn’t a recognition problem. We’ve recognized it. We’ve been recognizing it for decades. The problem is structural, and it survives recognition.

Most organizations don’t pay for prevention. They pay for survival.

That sentence is the whole essay. Everything else is unpacking why it’s true, and what it costs.

The default reading of the engineer story is “incentives matter.” This is true, but it explains nothing. Of course incentives matter. The question worth asking is why this same pattern keeps appearing across decades, across industries, across geographies — even in companies whose stated values explicitly elevate prevention.

The answer isn’t that performance systems are misaligned. It’s that performance systems can only reward what they can see.

And the boundary between what an organization can see and what it can’t isn’t a neutral measurement. It’s a creation event.

A bug that didn’t happen doesn’t generate a story. A fire that wasn’t lit doesn’t have a hero. A breach that was quietly prevented doesn’t make it into the board update. A risk that was structurally eliminated doesn’t show up on the dashboard.

Had 9/11 never happened, those preventing it wouldn’t have received any glory. Yet going after the terrorists that did it became global news for years.

Anything an organization can see, it can reward. Anything it can’t see, it functionally punishes by absence of any such reward. The prevention layer of every organization is, by construction, invisible.

Which means a rational employee who wants to be recognized has a clear strategic choice: do work the organization can see (that benefits them), or do work the organization can’t see (that quietly benefits the company).

Most of them, eventually, choose to be seen.

The ones who don’t, often leave.

The same pattern, in different costumes

You can find this structure almost anywhere you look inside a modern organization.

In engineering, the people who prevent incidents disappear into the background. The people who land the hot fix get promoted. Over time, the organization grows good at incident response and quietly worse at the disciplines that would have made the incident response unnecessary in the first place.

In risk and compliance, the function gets measured by the volume of documentation it produces, because documentation is visible. Whether the documentation actually changed the underlying risk posture is much harder to see. So risk becomes ceremonial. Compliance becomes theater. The team that quietly made certain problems impossible looks identical, from the outside, to a team that did nothing. I miss those “X Days Without an Incident” signs that used to get posted in work areas.

In AI governance, the same dynamic is being instrumented in real time. Most AI risk reviews are checkbox exercises because checkbox exercises are visible. The actual work — careful capability scoping, evaluation discipline, decision-rights design — produces no story. So AI governance teams produce policies. They produce frameworks. They produce decks. What they often don’t produce, because no one is paying for it, is fewer incidents.

And then there’s healthcare. Which is the example worth slowing down for.

A Western physician is paid when you’re sick. The visit, the procedure, the prescription, the follow-up — each one is a billable event. The healthy patient is essentially invisible to the revenue model. The doctor who keeps a patient well for thirty years produces far less measurable activity than the doctor who manages that patient’s chronic conditions across the same period.

Think about that for a moment.

The system isn’t necessarily “wrong”. Most doctors aren’t villains. The patients aren’t being failed by anyone in particular. The structure is simply incapable of seeing prevention as work. So the people inside it — doctors and patients alike — drift, year by year, toward the version of medicine the system can actually measure.

There is, in the tradition of ancient Chinese medicine, a different model. Ancient Chinese physicians were paid a retainer for as long as their patients remained well. When the patient became ill, the payments stopped, and resumed only when health was restored. A doctor who frequently resorted to surgery was considered an inferior doctor — the need for surgery was itself evidence that earlier, quieter work had been missed. The Chinese classics record a related line:

The sage does not treat those who are ill, but those who are well.

Whether that arrangement was ever as clean in practice as the legend suggests is a fair question. But I think the structural insight survives the historical scrutiny.

A system that pays for visible failure will produce more visible failure.

A system that pays for invisible success has to first learn to see invisible success. Which is its own discipline, but at least it’s pointed in the right direction.

The Western healthcare system has been recognizing this misalignment for decades. Like the Dilbert strip from last century, the recognition has not translated into change. Recognizing the misalignment is not enough. The structure has to be redesigned around what it chooses to make visible.

What this is actually about

The default explanation for this kind of dysfunction is that organizations have the wrong incentives, and the fix is to design better ones. That framing is true, but slightly off. It treats incentives as a layer you can tune, like volume on a stereo.

The deeper structure is that visibility is upstream of incentives.

Before you can reward something, you have to be able to see it. Before you can see it, your operating model has to be designed to surface it. And most operating models are designed to surface the wrong things. Not because their architects intended that, but because visible activity is easier to instrument than invisible prevention (squeaky wheel gets the oil and all that). And instrumenting the easy thing is what every system tends to do when no one is paying close enough attention to the harder thing.

This is why “fix the incentives” rarely fixes anything. The KPI gets adjusted. The bonus structure gets reweighted. The behavior shifts at the margin. And six quarters later, the same pattern reappears in a slightly different outfit, because the underlying visibility structure was never touched.

The boundary between what an organization sees and what it does not see isn’t a measurement. It’s a choice about what to bring into being.

This is the part most leaders miss. By choosing what to celebrate, an organization isn’t describing the work that exists. It’s determining what kind of work will exist next quarter.

Celebrate the firefighter, and the organization will produce people who are exceptionally good at fighting fires.

Celebrate the person who quietly designed the system so that fire was no longer possible, and the organization will produce people who quietly redesign systems to prevent fires.

Most organizations are very good at the first one.

Very few have ever seriously tried the second.

A different way of looking

If “fix the incentives” doesn’t work, what does? Not a recipe, but rather a different way of looking at your own organization, and at yourself. A small set of honest questions, the answers to which are usually much quieter than expected (so you should be sure to listen carefully).

What gets celebrated here? Not officially. Actually. Look at the all-hands stories. Look at whose name appears in the promotion announcements (and why). Look at what the founder mentions on the earnings call. That’s the visibility structure.

What never makes it into a story? If you can’t remember the last time someone was recognized for preventing an incident that never happened, your organization has structurally classified prevention as not-work.

Whose name gets attached to outcomes that didn’t happen? This is the hardest question, because the absence is the point. The people whose work prevented the bad outcome will, by definition, be harder to find than the people who responded to it. You may have to go looking for them. They won’t be in the room when the save is celebrated.

Where would you find the people doing quiet prevention work, and do they get promoted, ignored, or quit? Pay attention to the last one. The slow leaving of quiet prevention people is a leading indicator that the visibility structure has flipped permanently against them. By the time you notice the gap they leave behind, they’ve usually been gone for a while.

And the question almost no organization asks: What would you have to stop rewarding before prevention had room to be seen?

Every visibility structure has a fixed amount of attention. Recognizing prevention isn’t an additive change. It requires taking attention away from the heroics currently absorbing it. Which is uncomfortable, because the heroics are real, and the people doing them aren’t the problem.

The discipline isn’t “design better incentives.” It’s to notice what your visibility structure has been calling into being, and to decide whether that is what you actually wanted.

What disappears first

The structural cost of celebrating only what you can see is that the organization gradually optimizes itself toward producing visible things — including, eventually, the visible failures it then heroically survives.

The same crises keep happening. They keep being heroically survived. The heroes keep being rewarded.

And underneath all of it, slowly, the people who could have made the crises impossible are quietly leaving, or quietly waiting, or quietly storing the knowledge until it becomes a story they can tell.

What disappears first is the thing you were never rewarding to begin with. Leaders who only see the rescue have already paid for the wreck. They just haven’t read the invoice yet.

The harder discipline of leadership isn’t deciding what to reward. It’s noticing what was already happening that no one was rewarded to do, and recognizing that the boundary between what you elevate and what you ignore was never a neutral choice. It was always, quietly, a decision about what kind of organization to call into being.

The Dilbert strip is thirty years old. The ancient Chinese physicians are much older than that. The pattern has been visible for a very long time.

What has been missing is not the recognition.

It’s the willingness to look at your own organization — and at yourself — and ask the question that visibility structures are designed to make impossible to ask.

Who here is doing the important preventative work I can’t see?