The Decision Window Before Failure
This chapter examines the venture lifecycle around the moment when a startup should continue, change direction, or end. The focus is not chronology for its own sake, but the transitions that move resources, authority, and information between states. Seeing those transitions lets readers distinguish a deliberate pivot from a living-dead venture and understand why an apparently personal decision is shaped by the entrepreneurial ecosystem around it.
The lifecycle is a sequence of decisions, not just dates
A venture begins with an idea and a set of assumptions: someone has a problem, a proposed solution can address it, and enough people may value that solution to justify the resources required to build it. At the exploration stage, uncertainty is expected. The founder is not yet defending a proven business; they are deciding which assumptions deserve a test.
Testing changes the nature of the work. Instead of asking whether the idea sounds promising, the founder seeks evidence from customers, technology, operations, or other conditions that determine whether the venture can work. A test does not need to produce immediate success to be useful. It needs to change what the team should believe or do next. If the evidence strengthens the underlying hypothesis, the venture can proceed. If it exposes a problem that can be addressed by changing the business idea, the venture can learn and pivot. If it shows that the venture no longer has credible market or technological potential, the decision window for termination opens.
That last transition is often hidden because “failure” is treated as an event at the end of the story. The Amsterdam study uses a more consequential definition: failure can be the condition in which the venture no longer demonstrates viable potential or no longer justifies the founder’s continued investment of time and resources. Formal insolvency may come later, or not at all. A business can continue to pay some bills, attract another programme, or maintain a website while the case for continued founder effort has already weakened.
Each state carries something forward and loses something else. Exploration preserves optionality but has little evidence. Testing converts time and money into information, but it also makes some assumptions harder to defend. A pivot preserves the people, knowledge, and relationships assembled around the venture while changing its proposition. Decline reduces the value of those resources if it is not recognized, because more time can be spent protecting an old hypothesis instead of learning about a new one. Termination loses the current venture but can release founders, knowledge, talent, and attention for another use.
The transitions are not automatic milestones. Moving from exploration to testing means accepting that an idea must meet an external condition, not merely become more polished. Moving from testing to decline means recognizing that the evidence has weakened the venture’s claim on further investment. Moving from decline to termination requires more than noticing bad news; it requires an exit path that the founder can actually use. And moving from decline back to testing through a pivot requires a changed hypothesis whose result could prove the new direction wrong. These are decision transitions, not labels assigned after the fact.
This is why timing is a systems issue. The longer a non-viable venture persists, the more resources remain committed to it and the more difficult it can become to tell whether another effort is learning or merely preserving the past. Resource lock-in is not only a line item in a budget. It can be the founder’s working life, the team’s attention, the knowledge that is not being applied elsewhere, or the emotional capacity required to keep explaining why the next attempt will be different. The study describes this prolonged state as “living dead” persistence: a venture that neither grows through credible progress nor ends. The concern is not that every venture must move quickly. It is that a dynamic ecosystem needs a functioning route for resources to leave a venture that no longer justifies continued investment.
That route also has to distinguish a false alarm from a genuine decline. A venture can be early, imperfect, or temporarily under-resourced without being non-viable. The evidence question is therefore not “Did the startup encounter a setback?” but “What would have to change for continued investment to be justified, and has that condition become more plausible?” The Amsterdam study does not provide a numerical threshold for answering this question. It does provide a disciplined place to look: market or technological potential, the founder’s continued investment of time and resources, the signals produced by surrounding institutions, and the practical availability of termination.
The route from decline to termination is not necessarily a straight line. A founder may discover new evidence that changes the business idea’s core. The academic-entrepreneurship study used elsewhere in this series reports that core changes after negative feedback were positively associated with commercialization among 316 academic-led teams. That finding supports a useful distinction: a venture can remain alive because it is testing a materially revised hypothesis. It can also remain alive because each negative result is absorbed as a reason to try something else without ever changing the question of whether the venture is viable.
The observable activity may look similar in both cases. There may be a new product, a new customer segment, a new pitch, or a new funding request. A peripheral adjustment may improve delivery while leaving the central value proposition untouched; a core change may alter what problem is being solved, for whom, or how the venture expects to create value. The accessible evidence does not justify treating one kind of change as universally superior. It does show why the object of the change matters when we ask whether the team is learning.
The transition trigger is different in an adaptive pivot and in living-dead persistence. In an adaptive pivot, external information changes the venture’s logic and creates a test whose result could still justify stopping. In living-dead persistence, activity may increase while the underlying viability question remains unresolved. The difference is not movement versus stillness; it is whether movement changes the evidence-bearing hypothesis, makes the next judgment more informative, and preserves termination as a real option.
Why individual judgment is not the whole system
A founder’s decision is still part of this process. Prior investments of money, time, and employee effort can make abandonment painful. Emotional attachment can make criticism feel like an attack on identity rather than information about a venture. Optimism, overconfidence, persistence, and the desire to save face can all affect how a founder interprets bad news. These factors matter, but an explanation that stops there treats the decision environment as if it were neutral. It also makes the founder responsible for solving a problem whose information and power may be distributed across many institutions.
It is not neutral. The ecosystem affects what information reaches the founder, how credible it appears, what resources are available, who has authority, and what emotional costs are attached to admitting decline. A grant can be read as evidence that a respected institution believes in the venture. An investor’s encouragement can make another attempt seem responsible. A programme’s silence can make a founder assume that support is unavailable. A tightly connected network can make disclosure feel dangerous if bad news is likely to circulate. These are not just external circumstances surrounding a private judgment; they help determine which judgment seems reasonable at the time.
The same psychological tendency can therefore produce different outcomes in different systems. Sunk-cost thinking is harder to correct when institutions keep supplying fresh resources that appear to validate the original investment. Optimism is less obviously irrational when a powerful stakeholder insists that one more pivot could recover value. Reluctance to disclose problems is understandable when the available adviser has a financial interest in survival. A founder who hides declining results may be protecting status, access to capital, or a relationship on which the venture depends, even while that silence reduces the chance of receiving useful help. The founder’s cognition is not replaced by the ecosystem; it is formed and constrained through interaction with it.
This perspective also changes how feedback should be interpreted. Negative information is useful only if the founder can receive it without having to discount the source, and if acting on it does not immediately threaten their ability to continue or leave. A customer may reveal that the product is not solving a sufficiently important problem. An investor may respond by recommending more effort because their own return depends on survival. A peer with no stake may help the founder compare both signals. The content of the information is only one part of the decision; the relationship that carries it affects its meaning and usability.
This also changes how “agency” should be understood. Agency does not require a founder to possess unlimited freedom. A founder may make choices, but those choices can be shaped by a contract, a financial dependency, a threat of reputational damage, or the absence of a confidential person who can offer an independent view. Conversely, an ecosystem explanation should not turn founders into passive objects. The point is to identify the conditions under which their judgment can be informed, revised, and acted upon. A founder-centred system is not one that removes difficult decisions; it is one that makes the relevant evidence, alternatives, and consequences visible enough for the founder to own the decision.
In this sense, the lifecycle has both a technical and a social side. The technical side asks whether the venture can produce a viable market or technology. The social side asks who is allowed to define viability, whose evidence is trusted, and what happens when the answer is no. A startup can fail the first test while passing the second—institutions may continue to treat it as a promising venture. Or it can receive harsh social signals before the market question has been fairly tested. The ecosystem’s task is not to guarantee success; it is to prevent either kind of signal from silently replacing a careful decision.
The Amsterdam research is suited to this kind of explanation because it is designed to see both sides of the relationship. It draws on interviews with 30 founders whose ventures had failed and 25 stakeholders, alongside observations, documents, and a feedback event. The researchers used these sources to trace recurring patterns and contradictions rather than to calculate how common each mechanism is. The evidence is therefore strong for understanding how the process can work and how participants experienced it, but it is not a statistical estimate of all startups or all ecosystems.
That boundary matters when applying the lifecycle model. The model does not say that every venture in Amsterdam follows the same path, that every founder who persists is biased, or that rapid closure always maximizes value. It says that the transition from decline to termination is consequential and socially embedded. The case gives us mechanisms to inspect: who is sending a signal, who benefits from continuation, who bears the cost, and whether the founder can obtain advice that is both candid and usable.
A healthy lifecycle therefore requires more than a founder’s willingness to be realistic. It requires credible signals, room to change course, and a termination path that does not depend on pretending everything is fine. When those conditions exist, persistence can be disciplined experimentation. When they do not, persistence can become a way for the ecosystem to avoid resolving its own conflicting interests.
The next question is how those conflicts enter the decision environment. Startup culture can teach rapid testing while also treating perseverance as proof of character. Institutions can endorse a venture for reasons that have little to do with its market evidence. The lifecycle tells us where the decision belongs; the next chapter examines why the signals arriving at that decision can point in opposite directions.