When Is Stepping Back a Courageous Decision?

In an imagined scene, two partners sit before the prototype of a product they have worked on for months. The papers are orderly, the branding is beautiful, and the trial photographs resemble the dream they began with. But purchase orders are fewer than they expected, and the next batch requires new money. One says, “After all this effort, we cannot stop now.” The other falls silent, not because he has found the answer convincing, but because he remembers the nights they spent working together. At that moment, suggesting they stop seems like declaring all the preceding months worthless.

Yet the decision before them is not whether those months happened. They already did. It is whether the work ahead is worth the time, money, and responsibility it will demand. The distance between these questions is small in language but large in its consequences: do we want to improve the future, or force the future to vindicate the past?

This article does not urge entrepreneurs to withdraw at the first setback, nor does it recommend a particular investment. It examines how we think about continuing and stepping back, and the limits of the concept of “sunk cost.” Perseverance can be courageous, and stopping can be courageous too; naming either one is not enough to establish that it is the better decision.

Opening Questions

• When does what we paid before cease to be information and become pressure determining what we will pay next?
• How do we distinguish considered perseverance from continuing to defend our self-image?
• What is the difference between an unrecoverable cost and obligations that still stand?
• How can we step back fairly toward partners, employees, and those who trusted us?

1. The Past We Bring into a New Decision

In its economic sense, a sunk cost is an expense already incurred that can no longer be recovered. Not every past expense is entirely sunk: a machine that can be sold still has recoverable value, and a payment that can be partly refunded is not the same as one that is gone for good. The discussion should therefore begin by identifying what can actually be changed, rather than applying the term to everything that came before.

In a simplified numerical example we have devised, a business owner has paid sixty monetary units that cannot be recovered. Completing the project requires another twenty, and he expects to receive thirty from completion. We assume certainty and set aside all other alternatives and obligations solely to illustrate the idea. Comparing thirty with twenty makes completion preferable to stopping under these assumptions, even though the project as a whole will not recover the eighty. An overall project loss does not automatically mean that the next step is a bad one.

If the expected return were only ten against the new twenty, however, the previous sixty would not make this step any better. Both cases matter: excluding sunk costs from a forward-looking comparison does not amount to always recommending withdrawal. It may support completion in one case and stopping in another. These figures are invented for teaching purposes; they represent neither a guaranteed return nor an adequate model for evaluating a real project.

This idea is easily lost when we say, “We must recover what we paid.” Recovery is an understandable desire, but it needs a route that can be examined. Which customer will pay? For what value? What will it cost to reach them? Without an answer, the amount previously lost cannot do this work. Old money does not write a marketing plan or create new demand merely because we want it badly.

That does not make the past worthless. It may have left customer data, implementation experience, professional relationships, and perhaps assets that can be used in another direction. We need to separate these present benefits from the wish to justify what is over. Acquired knowledge is a possible reason to improve today's decision; the fact that an experience was painful is not, by itself, evidence that funding it again will put it right.

We can draw three columns on a sheet: what is over, what remains available to us, and what we will commit to now. Under the first, we place unrecoverable expenditure. Under the second, assets, skills, and data. Under the third, new options with their costs and risks. This is a proposed practical arrangement, not a standardized psychological test. Its usefulness lies in keeping three kinds of things from being mixed into the single sentence, “We have invested so much.”

2. What Does the Evidence Say—and What Does It Not Say?

The influence of money already paid on a decision may seem obvious, but intuition does not determine its size or the conditions under which it appears. A meta-analytic review by Stefan Roth, Thomas Robbert, and Lennart Straus, published online in 2014 and in a 2015 volume, brought together findings from earlier studies. It found evidence of a sunk-cost effect, with its strength and influencing factors varying by decision type. It distinguished between using something already paid for and allocating additional resources to an ongoing project. Its criteria were limited to particular contexts and monetary investments; it cannot legitimately be turned into a judgment about every human commitment or relationship. Original research.

This excerpt does not replace the next sentence in the paper, which qualifies the finding by decision type and its influencing factors. That is a methodological point: finding an effect across a collection of studies does not allow us to diagnose a particular person from afar, nor does it make every decision to continue a mistake. We first need to know the alternatives and information available, then examine the reason on which the decision rested.

In everyday reasoning, there is a difference between “I use what I bought because it still benefits me” and “I add new expenditure because I paid before.” The first may be entirely reasonable. Indeed, making use of what we already own may be better than buying a replacement. The problem begins when the old price becomes a sufficient argument even though the current option no longer serves its purpose, or when we ignore a new cost that could be avoided.

Consider an imagined example outside investment: someone enrolls in a workshop, then discovers that its level does not suit their needs. If they can transfer the registration or request a refund, those are current options. If neither is possible, the question of the time they will spend there remains. They may attend for a partial benefit or to accompany a friend, or choose another activity. The fee alone cannot establish whether the decision is sound; what matters is what they will gain now and what they will forgo.

Nor should we treat the names of biases as weapons for winning disagreements. Telling your partner, “You are committing a fallacy,” does not resolve the question of expectations. They may have information you have not heard, or may overestimate it. Ask them to explain how the new evidence changes the comparison between options. The discussion then moves from a test of their character to an examination of an argument that can be accepted or corrected.

3. When a Project Becomes a Defense of Its Owner

Return to the partners in the opening scene. One might say, “If we stop, what will we tell the people we said we would succeed?” This is a real social question, but it differs from whether the product can sell. The cost of embarrassment may not disappear merely because we recognize the difference. What we can do is make it visible instead of hiding it inside financial forecasts that appear objective.

One interpretation is that the person wants to preserve an image of being a careful decision-maker, and may feel that changing the plan calls their seriousness into question. This is a possibility for understanding, not a diagnosis or an experimental finding we attribute to every entrepreneur. If it is indeed how they feel, they can distinguish two statements: “I want my previous decision to be right” and “I want my next decision to be responsible.” The second does not deny the first, but identifies where their present agency lies.

There is also an ethical difference between acknowledging a mistake and degrading oneself. The first hypothesis may have been reasonable given the information available at the time, before the market changed or an important test failed. Or there may have been genuine shortcomings in the research that deserve acknowledgment and repair. In either case, we do not need to make the project succeed at any price in order to learn from it, or declare its owner incapable of every future decision.

The team can hold a meeting in which each person explains what they believed and what evidence changed that belief. One writes, “I thought interest expressed in the survey would become paid orders, and it did not.” Another writes, “I estimated production time below what the trial revealed.” These are specific formulations that make learning possible. “We were unlucky” or “Everything was wrong,” by contrast, may go beyond the available evidence.

It is important that the meeting not become a stage for blaming the member most willing to acknowledge a mistake. If candor brings personal punishment, people may learn to conceal problems rather than correct them. This is an organizational and ethical argument: anyone seeking honest information should make it possible to discuss it without humiliation. That does not remove accountability; it ties accountability to what the person did and was responsible for, rather than to their entire worth as a human being.

Conversely, “learning” must not become a cosmetic phrase that justifies any spending. What exactly did we learn? Can it be used? Is there a less costly way to test the next hypothesis? Without a specific answer, we may merely have replaced “We will succeed” with “We are learning” while leaving the decision without a standard. Learning deserves time and money, but naming it does not exempt us from specifying what it means.

4. A Serious Objection: Without Perseverance, No One Would Succeed

This objection deserves a hearing. Many undertakings do not produce results on the first attempt, and difficult work requires tolerating delay and adjustment. If every early loss becomes a reason to stop, we leave no room for a long-term project or a skill that needs practice. Excessive movement between ideas may also prevent the experience from accumulating that we need to make a judgment in the first place.

In this chapter, Epictetus discusses preparing for what an undertaking demands, using athletic competition and its requirements as an example. We do not present him as a behavioral economics researcher or derive a decision to close a project from his words. Their philosophical value here is to restore attention to expected costs before entering an undertaking, rather than rushing in and then being surprised by hardship.

But the appeal to perseverance does not, by itself, answer the question: what are we persevering in? We can remain committed to a purpose while changing the means. A team seeking to provide an educational service may discover that its lengthy program does not suit its audience, then try a shorter format or a different audience. Changing the format does not necessarily abandon the mission, just as retaining the first format is not sufficient evidence of loyalty to it.

The practical distinction proposed here is to look for a clear, future-oriented reason to continue. Have repeated paid orders appeared? Has delivery cost actually fallen? Has the earlier trial's difficulty been explained in a way that can be tested? These are examples of questions, not universal conditions every organization must satisfy. A research or artistic project may have aims that are not measured by sales, but it too needs to define what counts as progress toward its purpose.

Waiting can be a third option, provided we do not hide its cost. Freezing a project may preserve an opportunity, or it may consume rent and effort while leaving equipment idle. “Let us close temporarily” is not enough if temporary closure continues to drain the same resources. We need to specify what will remain, who will manage it, and when the situation will be reviewed. A pause is a decision requiring an explanation, just as full continuation does.

Likewise, “If we were starting today, we would not start” is a useful provocation, but not a final verdict. We are not starting from zero today: we have assets, a team, contracts, and information, which may make completion better than beginning a new alternative. The more precise question is: if we had to choose the next step on the basis of the entire current situation, rather than our wish to vindicate the past, what would we choose?

5. In the Light of the Quran and Islamic Wisdom

An ethical perspective adds the question of rights to the comparison between options. Stopping may cost the business owner less, but does not permit concealing information or leaving obligations to others unresolved. Nor does persistence become a religious virtue simply by being called determination. The texts considered here offer values of trustworthiness, consultation, honoring commitments, and responsible spending; they provide neither a ruling on a particular contract nor experimental proof of a psychological bias.

The verse brings together returning trusts to their owners and judging fairly. One ethical application we propose is that a manager include the rights of investors, employees, and customers in the review, rather than considering only what preserves their image. This is an application to management, not a claim that the verse specifies a financial model or ranks liquidation options. The details of rights require knowledge of the facts, contracts, and the relevant specialist authority.

Honoring commitments makes the question of exit a responsible one: what did we promise? What is the other party entitled to? Does the agreement allow termination or amendment? We cannot use a general text to answer questions about a contract we have not read, or turn sunk cost into a pretext for canceling a debt or a right. Existing rights are not past expenditure that can be erased from the calculation merely because they inconvenience the decision-maker.

The Prophet, peace and blessings be upon him, is addressed in a context of gentleness, pardon, and consultation. Al-Saadi explains consultation in matters requiring consideration and thought, then connects resolve with reliance on God after examining what needs consultation. Tafsir Al-Saadi. One educational reflection here is that firmness does not require rejecting a different opinion, and consultation does not mean postponing a decision forever. We do not claim that reliance on God guarantees a project's profit or that changing a plan indicates deficient faith.

This prohibition appears alongside the command to give relatives, the needy, and travelers their due. The value of money is therefore not confined to maximizing its owner's benefit, nor is responsible spending reduced to withholding generosity. In our discussion, this meaning can inspire a question about responsibility in using resources; it does not authorize calling every unsuccessful experiment wasteful. Judging an experiment requires considering its purpose, how it was evaluated, and the information available.

These values do not remove the pain of ending work one loves. A person can grieve for a dream while fulfilling others' rights. They may also choose to continue after consultation and a review of the reasons, making a decision consistent with their responsibility. The distinction is not between an optimist and a pessimist, but between a decision responsive to facts and rights and one that tries to silence them so the original story can remain unchanged.

6. From the Question of Withdrawal to a Method of Review

The following applications are editorial suggestions for organizing discussion, not an experimental program proven to eliminate bias. Use what suits the scale of the decision without turning every small choice into a lengthy file. The aim is to make our reasons visible and leave room to correct them before adding a new commitment.

Write the Next Decision at a Scale You Can Understand

Instead of “Should we continue the project?” write, “Should we order an additional batch this month?” Then specify what follows from yes and what follows from no. Continuing to serve current customers may be obligatory or useful, while expansion is a separate decision. Separating decisions sometimes reveals that the choice is not between saving everything and canceling everything.

Place the optimistic and pessimistic estimates alongside the most likely estimate, explaining the reason for each. Do not give numbers false precision; if you do not know a probability, say so. The next step may aim to gather a small piece of information before a larger commitment. In that case, you should know which question the experiment will answer, what resources are acceptable to complete it, and which decision its result will affect.

Test the Decision When Assumptions Change

Suppose, in another teaching example, that a team needs ten weeks to finish a trial service. It believes a potential customer will use it, but has not yet secured a clear agreement. “We are close to the finish” may sound like sufficient reason to continue, although it does not answer whether the service is needed. The task is not to deny work already done, but to identify what gives the coming weeks meaning: establishing the need, conducting a specific technical test, or fulfilling an existing obligation.

The team can compare three situations: the customer agrees, postpones, or declines. In each, it asks which parts of the product can be used, what additional work is needed, and what alternatives are available. If the decision looks good only with immediate agreement, that fragility should be visible in the discussion. If the service retains clear uses in all three situations, that is a different argument from clinging to elapsed time. The example assigns no probability to any situation; it reveals the effect of an assumption that may otherwise remain hidden.

The team then asks what information can be obtained at the lowest cost before completing the ten weeks. A careful interview may be enough to understand the need, or may be insufficient because the customer has not defined it either. A limited prototype testing one function may be needed. Neither an interview nor a trial is a guaranteed solution; its value depends on its ability to answer the actual question, not simply on calling it research or testing.

At the end of this review, the team may choose to continue. What matters is being able to say, “We will finish because this evidence has strengthened this hypothesis,” rather than “We will finish so the previous weeks were not wasted.” It may choose to complete a small part and then stop. Or the cost of the test itself may prove greater than its expected benefit. All are possible outcomes, which is why the meeting should not be organized to permit only confirmation of the idea owner's wishes.

Keep brief minutes stating what you know and do not know, the decision, and who is responsible for reviewing it. When the result appears, compare it with the written assumption, not a story rewritten after success or loss. This is a proposal for documentation and honesty in learning, not a promise to prevent mistakes. We may decide carefully and still not obtain the result we hope for, while becoming better able to identify what to change next time.

Separate the Project's Facts from Its Owner's Feelings

On one page, write what you know about demand, cost, and time. On another, write what you fear: disappointing a partner, public comment, or feeling that you have lost part of your identity. Do not treat the second page as a shameful thing to hide. But it needs language suited to it: conversation, support, and explanation, not invented sales figures concealing pain.

A founder might tell the partners, “It is hard for me to suggest stopping because the idea was mine, so I want us to review the assumptions together.” This is an imagined statement, but it illustrates how one can acknowledge personal interests without surrendering the decision to them. Acknowledgment does not automatically make a person neutral, but gives others a chance to understand the position and discuss evidence rather than guess at motives.

Give the Review a Date and a Standard

If you decide to give the project another chance, write down what you expect to learn and when. Do not make the date merely an occasion to move hope into the following month. Standards may change as circumstances change, but document the reason before counting the result as a success. The aim is not to punish a project for learning, but to keep the meaning of success from shifting whenever an unwelcome result appears.

Where possible within the team, it helps to separate the person who prepared the forecast from the person reviewing it. Give the reviewer the data and assumptions, not only the conclusion you want to hear. Do not assume that an outsider necessarily understands the work better. The proposed role is to ask about uncertainties and request reasons that can be examined; responsibility remains with the decision-makers.

Design an Exit That Preserves What Can Be Preserved

Stepping back is more than an announcement. It may require arranging handovers, documenting what the team learned, informing those affected, and examining options concerning assets and obligations. Some of these details are legal or financial and require the appropriate specialist. The general editorial principle is that freeing oneself from a project's burden should not suddenly transfer that burden to people less able to bear it.

Useful parts of the experience can be preserved without claiming that they compensate for every loss. A good operating manual, a more accurate understanding of a customer's need, or a respectful professional relationship may carry into the next undertaking. We do not need to invent a complete victory story to acknowledge these remains. We can honestly say: we lost in this respect, learned in another, and will change what we can.

Conclusion

Courage in stepping back is not the courage to close a door in any manner whatsoever, but to take responsibility for looking again. What we paid before deserves documentation, learning, and accountability, yet it does not alone determine what we should pay tomorrow. Review may lead to continuation with a stronger case, a limited adjustment, or a fair exit. We do not have to like the decision; we need to be able to explain its reasons without hiding present facts behind past pain.

Questions for Reflection

1. Which part of my next decision is justified by an expected benefit, and which by expenditure that is over?
2. What information, if it emerged, would genuinely make me change my plan?
3. Can I explain the reasons for continuing without saying, “After all we have paid”?
4. Who will be affected if I step back, and what does fairness toward them require?