← Back to Blog

The Hero Problem: Why Your Best Employee Is Your Biggest Risk

July 14, 2026 · Anthony Franco

The Hero Problem: Why Your Best Employee Is Your Biggest Risk

Every team has a Dave.

Dave is the one who just handles it. The Tuesday report goes out because Dave stays late Monday and runs the thing only Dave knows how to run. When the export breaks, Dave has a workaround. When finance asks a weird question, Dave already has the answer in a spreadsheet with forty tabs and no labels. You love Dave. You've told Dave he's a lifesaver, which is a strange thing to say to someone whose absence would, in fact, cost lives. Metaphorically. Probably.

Here's the uncomfortable part. Dave isn't your strongest asset. Dave is your single point of failure with a good attitude.

Heroism is invisible until it's absent

Call it the Hero Problem: one person quietly absorbs the org's complexity, and because they absorb it well, nobody sees the complexity at all. The work looks easy because Dave makes it look easy. So it never gets documented, staffed, or funded. Why fix a process that isn't visibly broken?

Then Dave takes a two-week vacation, and you discover the process was held together by Dave.

The mechanism is simple and a little cruel. Heroism registers as competence, not as risk. A missed deadline gets a postmortem. A deadline that gets hit through undocumented heroics gets a thank-you and a return to the queue. You reward the behavior that hides the danger. The better your hero performs, the deeper the dependency, and the less anyone notices until the day they're gone.

And people leave. They quit, they get promoted, they get hit by the proverbial bus. Your operational continuity should not depend on Dave's continued good health and job satisfaction. Yet for most teams, it does. There's a name for this in engineering: the bus factor. The number of people who'd have to disappear before the work stops. For a lot of critical work, that number is one.

AI turned every hero into a shadow factory

It used to be that Dave's knowledge lived in a spreadsheet. Bad, but findable. Someone could eventually reverse-engineer the tabs.

Now Dave has a personal automation. The sales rep wired up a scraping bot to pull leads. The analyst built a macro that reconciles three systems before breakfast. Someone on ops is quietly running a private script against an API with their own login and a credit card they expense as "software." None of it is sanctioned. All of it works. This is shadow automation: individual acts of heroism, built with a personal tool and a personal account, invisible to everyone until the account gets deactivated.

It makes the Hero Problem worse in a specific way. Automation fails silently. A human hero, on a bad day, tells you they're drowning. A script doesn't. It runs, produces output that looks right, and quietly drifts wrong for weeks because the source page changed its layout in March. Errors compound across a hundred runs before anyone notices the numbers stopped adding up. You didn't just lose Dave. You lost Dave and inherited a black box that lies politely.

So the AI version of the Hero Problem is two failures stacked. The knowledge is gone, and the thing that replaced the knowledge is unaccountable.

Your job isn't to find more heroes

This is where leaders reach for the wrong fix. They try to clone Dave. Hire another Dave, groom a backup Dave, beg Dave to write it all down before he leaves (he won't; the knowledge is muscle memory now, and documentation is somebody else's Tuesday).

The goal isn't more heroes. It's less heroism.

That's not a demotion for Dave. It's a promotion for the work. You want the thing in Dave's head to become something the team owns and can see:

  • Make the invisible visible. Sit with your heroes and ask what they handle that isn't written down anywhere. The answer is usually longer and scarier than you expect.
  • Move it from person to system. A documented process, a shared and monitored automation, a runbook someone else can follow. Ownership by the team, not by the individual.
  • Watch the automations you can't see. If work is getting done and you can't point to who owns the tool doing it, that's not efficiency. That's a liability with good uptime.
  • Reward the boring version. Make "I made my job repeatable by someone else" worth more than "I saved the day again." Right now, most orgs pay for the second one.

The quiet reward here is that Dave gets to stop being the load-bearing wall. Heroes burn out precisely because the org made them irreplaceable and then acted surprised when they wanted a weekend. Free Dave to do the work only Dave can do, instead of the work only Dave happens to do.

Find your heroes. Thank them sincerely. Then get to work making them unnecessary, which is the highest compliment you can pay a hero and the one they'll resent the most.