Field Notes · Money

Zombie Projects

6 min read · organizational behavior

Nobody approved the waste on purpose. That's exactly why it's still there.

Every company past a certain size is running at least one initiative that everyone privately agrees should be shut down: the platform migration that stalled two-thirds finished, the analytics subscription three people still get an email about, the "Q2 strategic initiative" that quietly became a standing meeting with no output. Nobody defends these projects out loud anymore. And yet the invoice keeps clearing every month. Understanding why requires looking less at the projects and more at the people deciding, or not deciding, their fate.

The trap of money already spent

In 1985, psychologists Hal Arkes and Catherine Blumer ran an experiment that has become a cornerstone of behavioral economics. People who had already paid for a ski trip were more likely to go on it, even after learning of a better trip they'd rather take, simply because they'd already spent the money on the first one. This is the sunk cost fallacy: treating money, time, or effort already gone as a reason to keep going, when the only rational question is what happens from here.

Organizations run this experiment on a much larger scale, with much larger numbers, and far less scrutiny. A $400,000 platform migration that's 60% complete doesn't get evaluated on whether the remaining 40% is worth funding. It gets evaluated on how bad it would look to have spent $400,000 on nothing. That framing makes finishing feel inevitable, even when finishing is the more expensive path.

Why the original sponsor is the worst person to pull the plug

Barry Staw's foundational 1976 research on escalation of commitment found something sharper than the sunk cost fallacy alone: people escalate their commitment to a failing course of action most strongly when they personally chose it. The CEO who greenlit the CRM overhaul three years ago isn't just weighing the project on its merits. She's weighing it against what killing it says about her own judgment. That's not a character flaw. It's a predictable, well-documented feature of how humans relate to their own past decisions, and it means the person with the authority to end a bad project is very often the person structurally least able to see it clearly.

Killing a project doesn't feel like saving money. It feels like admitting a loss. Those are not the same decision, even when they're the same decision.

Loss aversion, and why quiet bleeding beats a clean cut

Daniel Kahneman and Amos Tversky's prospect theory (1979) showed that losses register roughly twice as painfully as equivalent gains feel good. Formally closing a project means booking a loss, a specific number, in a specific meeting, with your name near it. Letting it quietly continue costs more in total, but the cost is diffuse, deferred, and never has to be said out loud in one sentence. Given the choice between a small, certain, named loss today and a larger, vague, unnamed cost spread across the next eighteen months, loss-averse decision-makers reliably pick the second option. It isn't laziness. It's the math of how loss feels, working exactly as documented.

Slack that nobody's job it is to notice

Richard Cyert and James March's 1963 A Behavioral Theory of the Firm introduced the concept of organizational slack: the buffers, redundancies, and underused resources that accumulate inside any company that isn't in crisis. Slack isn't inherently bad; some of it is healthy margin. But slack that goes unexamined during good years doesn't get pruned until a much more painful, much more public cut becomes unavoidable. The three-person, hand-maintained process that could be a script isn't surviving because anyone decided it should. It's surviving because auditing it has never been anyone's whole job, only a piece of several people's, which functions exactly like nobody's, for the same reason unowned projects never start.

What actually fixes it

The research points to a structural answer, not a motivational one: the person who ends a wasteful project needs to be someone without a reputational stake in its original approval. That's a specific, nameable qualification, and it's one a long-tenured executive or a founder structurally cannot hold for their own decisions. It's also, not coincidentally, most of the job.