The sunk cost fallacy, and how to beat it

A few years ago I kept a strong engineer on a doomed integration for three months. We had budgeted one. By the end of week four it was clear the vendor API could not do the one thing the whole feature depended on, and no amount of clever code on our side was going to change that. I knew this. My engineer knew this. And I still could not bring myself to kill it, because we had already spent a month, and killing it meant that month was gone for nothing. So I asked for two more weeks. Then two more. We shipped a hollowed-out version nobody used, and then quietly retired it a quarter later.

The honest accounting is brutal. The right move at week four was to stop. Everything I spent after week four, I spent to protect a month that was already gone no matter what I did. That is the sunk cost fallacy, and once you learn to see it you will notice it everywhere, most uncomfortably in your own decisions.

This piece is about that one bias specifically. What the sunk cost fallacy actually is, the shapes it takes at work and in ordinary life, why an otherwise sharp brain keeps falling for it, and a small set of moves that let you walk away from what is already lost without pretending the loss did not happen.

What the sunk cost fallacy actually is

A sunk cost is any time, money, or effort you have already spent and cannot get back. The fallacy is letting that unrecoverable spend drive a decision about the future, when the only thing a good decision should weigh is what happens from here. Rational choice only cares about costs and benefits still in front of you. The past is fixed. It should be an input to what you learned, not to what you do next.

The tell is a specific sentence, and you have said it. "We have come too far to quit now." "I have already put two years into this degree." "I paid for the ticket, so I have to go even though I feel awful." Notice that none of those sentences is an argument about the future. Every one of them is an argument about the past, dressed up as a reason to keep going. The money is already spent whether you go to the concert or stay home. The only real question is which of those two evenings is better from right now, and the ticket price has nothing to do with the answer.

The money is already gone. The only question worth asking is which path is better from here, and what you already spent is not on that list.

Economists call the clean version of this rule "ignore sunk costs," and it sounds obvious written down. In the moment it is anything but, because the spend does not feel like the past. It feels like a debt the future owes you, and quitting feels like defaulting on it. That feeling is the whole problem, and it is worth understanding where it comes from.

Why your brain keeps falling for it

The sunk cost fallacy is not stupidity and it is not weak will. It is the output of a few specific mental habits that are useful most of the time and quietly betray you here. Name them and they lose some of their grip.

Losses hurt about twice as much as gains feel good

Daniel Kahneman and Amos Tversky showed that we do not weigh gains and losses evenly. A loss lands roughly twice as hard as an equivalent gain, a pattern they called loss aversion. Walking away from a project turns a fuzzy, unrealized cost into a sharp, official one. As long as you keep going, you have not "lost" the month yet, not really, not in a way you have to admit. Quitting is the moment the loss becomes real and gets its name. So you keep going, not because continuing is smart, but because it postpones the exact feeling your brain is wired to avoid most.

You want to look consistent, to yourself most of all

People have a deep pull toward acting consistently with what they have already done. Robert Cialdini calls it commitment and consistency, and it is usually a fine instinct. It makes you keep promises and finish things. But it also means that every dollar and hour you sink into a path is another public and private vote for that path being right. Quitting is not just abandoning a project. It feels like admitting that the last three months of you were wrong, and the self does not like casting that vote. So it rationalizes. It finds a reason the next two weeks will be different.

Quitting feels like a verdict on you

This is the one that had me by the throat with that integration. Stopping did not feel like a clean strategic call. It felt like standing up in a room and saying "I was wrong, and I spent the company's month proving it." Continuing let me keep that verdict pending. The bias thrives on the fact that an open project has not officially failed yet, in the same way an unmade decision cannot be wrong yet. If you have read our piece on how to stop overthinking decisions, you will recognize the shape: both are ways of keeping a painful truth in a state of "not yet," and paying rent the whole time for the privilege.

What it looks like in the wild

The fallacy is easiest to fight when you can recognize its costume. Here are the ones I see most, at work and away from it.

The doomed feature. A team is six weeks into an eight-week build when the market shifts and the thing is clearly no longer worth shipping. They ship it anyway, because stopping at six weeks feels like admitting six weeks of waste, while finishing at least produces something. The something is used by no one. They spent two extra weeks to avoid feeling the first six.

The hire who is not working out. A manager knows by month two that a hire is a poor fit, but keeps investing coaching, projects, and patience deep into month eight, partly out of care and partly because letting go means admitting the hiring decision, the onboarding, and all that coaching were spent on the wrong person. Every month of delay is more sunk cost making the eventual call harder, not easier.

The rewrite that will not die. Engineering leaders know this one intimately. A big rewrite is a year in and visibly behind, but there is now a year of work that only pays off if you finish, so you keep pouring people into it long past the point where the honest move is to cut scope or stop. I get into how to make the rewrite-versus-refactor call itself in the guide to improving your decision-making; the sunk cost trap is what keeps you from revisiting the call once you have made it.

And the everyday ones, which are the same bug in a smaller body. The half-finished book you are slogging through because you are already two hundred pages in. The membership you keep paying for so it does not feel wasted. The relationship or the city or the career that stopped fitting years ago, propped up by the sheer weight of the time already in it. Nations do it too. Wars and megaprojects get called too big to abandon precisely when abandoning them is the sanest thing left, which is why the fallacy sometimes goes by the grim nickname of the Concorde effect, after a beautiful airplane that everyone kept funding long after the math said stop.

How to actually beat it

Knowing the fallacy does not dissolve it. What helps is a handful of questions and habits you can reach for in the moment the pull starts, each one designed to yank your attention off the past and back onto the only thing that can still change: what happens next.

Ask the blank-slate question

This is the single most useful move, and it is one question. "Knowing what I know now, if I were starting fresh today with none of this already spent, would I choose to begin this?" If the answer is no, you have your answer, and the months already gone are not a reason to continue. They are the tuition you paid for knowing the answer is no. A colleague of mine phrases it as the newcomer test: what would a smart person who just walked in, with no history and no ego in the project, tell you to do? They would tell you to stop, and the only reason you cannot hear it is that you are not new.

Set the kill criteria before you start, not during

The cleanest defense is to decide the exit before you are emotionally underwater. Before you begin, write down what would have to be true for you to stop. "If we cannot confirm the API supports X by week two, we kill it." "If this hire is not clearly on track by month three, we part ways." Written in advance, in cold blood, these tripwires are a message from the version of you that had no sunk costs yet to the version that will be drowning in them. When you hit one, you are not quitting in the heat of a bad week. You are honoring a call you already made when you could think clearly.

Separate the decision from the confession

Part of what keeps you going is the dread that stopping means announcing failure. So split the two. The strategic decision is "we stop putting resources here." The story you tell about it is a separate thing you get to write well: "we ran a real experiment, we learned the vendor cannot support this, and we are redeploying the team before we waste another month." That is not spin. It is the truth, and it happens to be the truth that a good team tells about a good decision. Stopping early is not the failure. Continuing after you knew is.

Reframe the spend as tuition, not debt

Your brain files sunk cost as a debt that the future has to repay by making the effort "worth it." Refile it. It was tuition, and the lesson is already bought. The three months I spent on that integration taught the team exactly which vendor assumptions to test first next time, and that knowledge did not require a fourth month to become real. When you treat past spend as a completed purchase of information rather than a loan you have to work off, walking away stops feeling like default and starts feeling like graduation.

Watch for the reinforcements the bias sends

The fallacy rarely shows up alone. It brings confirmation bias, which makes you notice every faint sign the project might still work and wave off the loud signs it will not. It brings optimism about the very next stretch, the "just two more weeks" that has already broken its promise twice. When you catch yourself building a hopeful case for continuing, ask whether you are reasoning toward the truth or toward the conclusion that lets you avoid the loss. The answer is usually written in how hard you are working to believe it.

A short script for when you feel the pull

The next time you catch yourself saying some version of "but we have already put so much into this," run this. It takes a minute for small things and scales cleanly to the large ones.

1. 1Name the sunk cost out loud and admit it is gone. "We spent a month. That month is spent no matter what I choose now." Saying it drains some of its power to hide.
2. 2Ask the blank-slate question: starting fresh today, with everything you now know and none of this already spent, would you begin this? Answer honestly before you argue with the answer.
3. 3List only the costs and benefits still ahead. What will the next stretch cost, and what will it realistically return? The past does not go on this list.
4. 4Check whether you are reasoning toward truth or toward relief. If your case for continuing leans on the effort already spent, that is the fallacy talking, not a reason.
5. 5Decide from the future-facing numbers, then write the honest story of the call: what you learned, and where the freed-up resources go now.

The blank-slate question is doing the heavy lifting there. It is a small piece of mental surgery that cuts the past away from the decision and leaves you looking only at what you can still affect. Almost nobody asks it in the moment, because the moment is exactly when the loss feels most like a debt. That is why it works better as a habit than as a good intention.

Why reading this will not be enough

Here is the catch, and it is the honest one. You can finish this article, nod along, fully agree that sunk costs should be ignored, and then keep funding your own doomed integration next quarter. I did, and I already knew all of this at the time. The sunk cost fallacy is not a gap in your knowledge. It is a feeling, loss aversion and the fear of looking wrong, and feelings do not update because you read a well-argued paragraph. They update when you practice the new move enough times that walking away stops feeling like defeat and starts feeling like the ordinary, unremarkable thing a good decider does.

That practice is hard to come by in real life, because real bets pay out slowly and tangled up with luck. You kill a project, and you never get to run the other timeline where you did not, so you never quite feel the win of having stopped. You just carry the faint worry that maybe two more weeks would have done it. That missing feedback is exactly what teaches the bias to survive, and it is the loop most of us never get to close.

Find one thing to quit this week

You do not beat the sunk cost fallacy by resolving to be more rational, the same way you do not get fit by resolving to be stronger. You beat it one honest exit at a time. So find one thing on your plate right now that you are only continuing because of what you have already put in. A project, a subscription, a book, a commitment that stopped earning its keep a while ago and coasts on momentum.

Ask it the blank-slate question. If the answer is no, stop, and say plainly what the effort taught you on the way out. It will feel slightly wrong the first few times, like you are wasting something. That feeling is the fallacy, and it is exactly what you are training against. Do it enough and it flips. You start to feel the pull of "we have come too far" before it grabs you, you let go of the lost causes early, and you spend what you have on the paths that are still worth spending it on. That, and not stubbornness, is what finishing well actually looks like.