Most property companies running software they have outgrown know it. The system is slow, the workarounds have multiplied, the team complains, and the reporting no longer answers the questions leadership asks. And yet the decision to leave keeps getting deferred, for a reason that sounds responsible every time it is said aloud: we have put too much into this to walk away now. The years of data, the money spent on implementation, the training, the customization, the sheer effort of getting the organization onto this system, all of it feels like a reason to stay.

It is not a reason to stay. It is a reason the decision feels hard, which is a different thing. Almost everything in that list is already spent and cannot be recovered whether you stay or go, and the part that is not, the cost of switching, is a one-time payment weighed against a cost of staying that repeats every single year you remain. The instinct that treats past investment as a reason to continue is one of the most studied errors in human decision-making, and in software it keeps companies married to systems that are actively holding them back, long past the point where leaving was the obvious call.

The trap has a name, and a lot of evidence behind it

This is not a motivational observation. It is a well-documented cognitive bias with decades of research behind it.

The sunk cost fallacy is the tendency to continue an endeavor because of what you have already invested in it, even when continuing is no longer the best choice. The definitive study is Hal Arkes and Catherine Blumer's 1985 paper "The Psychology of Sunk Cost," which showed that people are far more likely to keep going with something once they have sunk money, time, or effort into it, regardless of whether continuing actually serves them. In one of their experiments, a large majority chose to keep funding a doomed project when the prior investment was mentioned, while almost nobody chose to invest in the same doomed project when it was not, the only thing that changed was the presence of a sunk cost, and it flipped the decision.

The rational rule the research establishes is stark: sunk costs should be ignored entirely. Only the incremental, forward-looking costs and benefits of your current options should affect the choice. What you already spent is gone in every version of the future, so it carries no information about which future is better. Arkes and Blumer found the psychological engine behind the error, too: the driver is the desire not to appear wasteful. Leaving the old system feels like admitting the investment in it was wasted, and to avoid that feeling, people keep pouring resources into the very thing that is failing them.

There is a related finding that stings even more in a software decision. The organizational researcher Barry Staw showed that people escalate their commitment far more to projects they themselves chose than to ones they inherited, because their own judgment is now on the line. The person most likely to defend the failing system is often the person who selected it, not out of stubbornness, but because leaving means conceding that the original choice no longer holds. This is escalation of commitment, the documented pattern of increasing investment in a failing course of action to justify what came before, and it is exactly the dynamic that keeps a company on software everyone knows it should leave.

Separate what is sunk from what is not

The way out of the trap is to sort the "we've put too much in" feeling into its two very different parts, because they belong on opposite sides of the decision.

Almost all of what makes leaving feel wasteful is genuinely sunk, and genuinely irrelevant to the choice in front of you. The money spent implementing the current system is gone whether you stay or leave. The years of using it are gone. The training, the customization, the effort of adoption, all spent, all unrecoverable, all completely unaffected by what you decide next. These costs feel like reasons to stay, but they are identical in both futures, so they cannot favor either one. Ignoring them is not reckless; it is exactly what a sound decision requires.

The switching cost is the one part that is real and forward-looking, and it deserves to be in the decision, honestly and fully. Migrating your data, retraining your team, the disruption of the transition, the temporary dip in productivity, these are genuine future costs you would pay by leaving and avoid by staying, and pretending they are trivial is its own mistake. But notice what the switching cost actually is: a one-time payment. You pay it once, in the transition, and then it is done. It is not a recurring reason to stay; it is a finite toll on the way out. That toll is real, and choosing to pay it badly, or at the wrong moment, is its own error, which is exactly the caution examined in why the best time to replatform is rarely now: the honest version of this argument respects the switching cost rather than waving it away.

So the honest decision is not "everything we've put in versus the cost of leaving." It is a much narrower comparison: the one-time cost of switching, against the ongoing cost of staying on a system that no longer fits. Everything else in the "we've invested too much" feeling is sunk, and belongs nowhere in the math.

The cost of staying is the one that compounds

Here is the asymmetry that the sunk cost feeling hides completely: the switching cost is paid once, but the cost of staying on the wrong system is paid every year, and it grows.

A system you have outgrown is not a fixed, tolerable burden. It is a compounding one. Every month on it is more staff time lost to workarounds, more errors from manual processes it forces, more decisions made on reporting that cannot answer the question, more opportunities missed because the system cannot support them, and more data accumulated that will eventually have to be migrated anyway, making the switch you are avoiding a little harder each year you avoid it. The cost of staying is not a one-time payment you have already made. It is a meter running, and it runs faster the longer you leave it.

Set the two against each other honestly and the picture usually inverts. The switching cost, painful as it is, is a known, one-time, finite number. The cost of staying is a recurring drain with no end date, and often a rising one. A company that stays to avoid the one-time cost is choosing to pay the recurring cost indefinitely, which is almost never the better deal once you actually compare them rather than letting the sunk investment do the deciding. The question was never "can we afford to switch." It was "can we afford to keep paying the cost of not switching, forever," and that question answers itself far more often than companies let it.

The honest part

Several qualifications keep this from becoming a reckless argument that you should always leave, which would be as wrong as never leaving.

Switching costs are real and legitimately belong in the decision, and this is not a case for treating them as nothing. The research is precise on this point: switching and wind-down costs are future costs, not sunk ones, so unlike your past investment they genuinely count. A migration can be expensive, risky, and disruptive, and a company that switches carelessly, underestimating the transition or moving to a system no better than the one it left, has made a real mistake. The argument is not that switching costs do not matter. It is that they should be weighed as the one-time future cost they are, against the recurring cost of staying, rather than inflated by all the sunk investment that has nothing to do with the choice.

It is also true that not every complaint about a system means it should be replaced, and sometimes staying is genuinely right. A system that is imperfect but adequate, where the real cost of staying is modest and the switch would be genuinely disruptive, may well be worth keeping, that is a legitimate outcome of the honest comparison, and it is closely related to the one-way-door caution in the systems decision you can't easily reverse. The point is to make the decision on the real numbers, the one-time switching cost versus the ongoing cost of staying, rather than on the sunk investment. Sometimes that comparison says stay. The error is letting "we've put too much in" decide before the real comparison is even made.

And leaving is genuinely hard in ways beyond the money, so acknowledging the difficulty is not the same as succumbing to the fallacy. There is real organizational fatigue, real risk, real disruption in a major system change, and a leader weighing those seriously is being responsible. The distinction is between weighing the genuine forward-looking costs of switching, which is sound, and being held in place by the feeling that past investment must not be wasted, which is the fallacy. One is diligence. The other is the trap wearing diligence's clothes.

Decide forward, not backward

The practical discipline is to strip the sunk investment out of the decision entirely and compare only what actually differs between your two possible futures.

Make the comparison explicit and forward-looking. On one side, put the honest, full, one-time cost of switching: migration, retraining, disruption, the productivity dip, all of it, not minimized. On the other side, put the ongoing annual cost of staying on the current system: the lost time, the errors, the missed opportunities, the reporting gaps, projected forward over the years you would keep it. Then compare those two numbers, and deliberately exclude everything you have already spent, because it appears identically in both futures and so cannot favor either. If the recurring cost of staying, compounded over a realistic horizon, exceeds the one-time cost of leaving, the sunk investment is not a reason to stay; it is just the feeling that makes the right decision uncomfortable.

There is a single question that cuts the fallacy out cleanly. If I were arriving at this company today, with no history and nothing already invested, and I saw the current system and the cost to switch, would I choose to stay on it? A newcomer carries no sunk cost and no need to justify a past choice, so their answer is the forward-looking one, uncontaminated by what has already been spent. If the honest answer is that a clear-eyed newcomer would switch, then you should switch, and the only thing keeping you is an investment that is already gone whether you stay or not. What you spent getting here cannot be recovered by staying here. It can only keep costing you, and the cost of switching, whatever it is, you pay just once.

FAQs

Q1. Isn't all the money and effort we've put into our current system a reason to keep it?
No. Almost all of that, implementation cost, years of use, training, customization, is sunk: already spent and unrecoverable whether you stay or leave, so it is identical in both futures and cannot favor either. The sunk cost fallacy is precisely the tendency to treat past investment as a reason to continue. What you already spent getting onto the system cannot be recovered by remaining on it, so it should carry no weight in the decision about what to do next.

Q2. What is the sunk cost fallacy?
It is a well-documented cognitive bias, the tendency to continue something because of what you have already invested, even when continuing is no longer the best choice. Arkes and Blumer's 1985 research demonstrated it experimentally: people kept funding a doomed project far more often when prior investment was mentioned than when it was not. The rational rule is that sunk costs should be ignored, and only the forward-looking costs and benefits of your current options should drive the decision.

Q3. So switching costs don't matter?
They do, and this is the key distinction. Switching costs, migration, retraining, disruption, are future costs, not sunk ones, so unlike your past investment they genuinely belong in the decision. The point is not to ignore switching costs but to see them for what they are: a one-time, finite payment, to be weighed against the recurring cost of staying, rather than inflated by all the past investment that has no bearing on the choice.

Q4. Why is the cost of staying worse than it looks?
Because it compounds. A one-time switching cost is paid once and done, but the cost of staying on an outgrown system, lost staff time, errors, missed opportunities, reporting gaps, recurs every year and often grows. Data keeps accumulating that will have to be migrated eventually anyway, making the switch harder the longer you wait. Staying to avoid a one-time cost means choosing to pay a recurring cost indefinitely, which rarely turns out to be the better deal.

Q5. Why is the person who chose the system often the one defending it?
Because of escalation of commitment, documented by Barry Staw: people escalate commitment far more to decisions they made themselves than to ones they inherited, since their own judgment is on the line. Leaving the system can feel like admitting the original choice no longer holds, so the selector defends it, not from stubbornness but from the discomfort of conceding. Recognizing this dynamic helps separate the decision from the ego attached to it.

Q6. Does this mean we should always switch?
No. Sometimes staying is genuinely the right call, when a system is imperfect but adequate, the real cost of staying is modest, and the switch would be genuinely disruptive. That is a legitimate result of the honest comparison. The argument is only that the decision should rest on the real forward-looking numbers, the one-time switching cost versus the ongoing cost of staying, rather than on sunk investment. Sometimes that comparison says stay; the error is letting past investment decide first.

Q7. How do we weigh switching costs honestly then?
Put the full, one-time cost of switching, migration, retraining, disruption, the productivity dip, unminimized, on one side. Put the ongoing annual cost of staying, projected over a realistic horizon, on the other. Compare those two, and deliberately exclude everything already spent, because it appears identically in both futures. If the compounded cost of staying exceeds the one-time cost of leaving, the sunk investment is not a reason to stay; it is just what makes the right decision feel uncomfortable.

Q8. What single question helps most?
Ask what you would choose if you arrived at the company today with no history and nothing already invested, seeing only the current system and the cost to switch. A newcomer carries no sunk cost and no past choice to justify, so their answer is the clean, forward-looking one. If a clear-eyed newcomer would switch, you should switch, and the only thing holding you is an investment that is already gone regardless of what you decide.