Decisions Under Ambiguity: How to Act When the Data Doesn’t Exist Yet

When the data you need has not yet been created, the standard decision-making toolkit fails. Ambiguity—the condition in which no reliable probability distribution exists—is not a temporary inconvenience for founders; it is the permanent operating environment. The practical question is not how to eliminate it, but how to act well inside it.

Why the standard toolkit fails founders

The dominant framework taught in business schools is expected utility theory (EUT): list all possible outcomes, assign probabilities to each, and choose the option with the highest expected return.¹ It is a coherent framework. It is also largely useless at the frontier of a new venture, because it assumes the decision-maker can enumerate outcomes and assign meaningful probabilities to them. In genuinely novel situations, that assumption collapses.

The distinction that matters here was drawn a century ago by economist Frank H. Knight. Knight is commonly credited with defining the distinction between decisions under “risk”—known chance, or measurable probability—and decisions under “uncertainty,” where probability is unmeasurable or indeterminable.² Knight proposed that this distinction was important for economic theory, because uncertainty affords opportunities for profit that do not exist in situations where risks can be calculated. That observation is not merely academic. Every genuinely new market is, by definition, a Knightian uncertainty problem—and founders are the people who walk into it anyway.

Daniel Ellsberg sharpened the point in 1961. Ellsberg popularized his paradox in the paper “Risk, Ambiguity, and the Savage Axioms,” and it is generally taken to be evidence of ambiguity aversion—the tendency to prefer choices with quantifiable risks over those with unknown, incalculable risks.³ He labelled this behaviour ambiguity aversion, noting that such an anomaly would violate the basic assumptions of subjective expected utility theory. In the fifty-plus years following Ellsberg’s conjecture, a large volume of empirical research has offered evidence that Ellsberg was indeed correct—the majority of people appears to exhibit ambiguity aversion. The implication for founders is uncomfortable: the human default, even among sophisticated operators, is to flee ambiguity rather than reason through it. Recognising that reflex is the first act of discipline.

How the mind shortcuts—and where it goes wrong

When founders cannot rely on clear probabilities, they fall back on mental shortcuts. Based on experiments carried out with volunteers, Tversky and Kahneman discovered that humans make predictable errors of judgement when forced to deal with ambiguous evidence or make challenging decisions. These errors stem from “heuristics” and “biases”—mental shortcuts and assumptions that allow swift, automatic decisions, often usefully and correctly, but occasionally to our detriment.⁴

Tversky and Kahneman described three heuristics employed in making judgements under uncertainty: representativeness, which is usually employed when people judge the probability that an object or event belongs to a class or process; availability of instances or scenarios, often employed when people assess the frequency of a class or the plausibility of a particular development; and adjustment from an anchor, usually employed in numerical prediction when a relevant value is available. These heuristics are highly economical and usually effective, but they lead to systematic and predictable errors. A better understanding of these heuristics and of the biases to which they lead could improve judgements and decisions in situations of uncertainty.

The practical danger for founders is specific. When entrepreneurs cannot rely on clear probabilities, they often fall back on mental shortcuts to make decisions. These behaviours are captured by models like random support theory, which explains how people estimate probabilities using a small number of salient examples or cues, often ignoring broader statistical patterns. Another, case-based reasoning, shows how decisions are shaped by past experiences—real or imagined—that may not actually apply. These shortcuts are common, but they often backfire.⁵ Analysis of startup evaluation panels in Canada found that reviewers frequently overestimated the chances of success for early-stage ideas, particularly when uncertainty was high. Their predictions often failed to match what actually happened—a sign that even experts struggle to judge accurately without the right tools or feedback.

Canonical work by Tversky and Kahneman established that people often seek to reduce uncertainty, sometimes by substituting heuristic judgements for more complex reasoning. A drive to reduce uncertainty can lead to unwarranted expressions of certainty, which has consequences for decision-making individually and at an organisational level. The founder who mistakes confidence for clarity is not making a bold call—they are making an unexamined one.

What research says expert founders actually do

The most rigorous alternative framework for entrepreneurial decision-making under ambiguity is effectuation, developed by Saras Sarasvathy at the University of Virginia’s Darden School of Business. Sarasvathy discovered effectuation in the late 1990s as a result of in-depth studies of how expert entrepreneurs think, act, and make decisions in the initial stages of new venture creation.⁶

Sarasvathy drew a distinction between the traditional “causal” reasoning—typical of managers who use historical data to predict outcomes in the environment—and the “effectual” reasoning that minimises reliance on prediction and is used by expert entrepreneurs to deal with uncertainty. The difference is not temperamental; it is structural. Effectuation is a decision-making process that commences not with a predefined goal, but with a set of means as given. It encourages entrepreneurs to harness their existing resources and to select from a range of possible effects that can be generated from these means.

The five principles of effectuation are worth stating precisely, because they are often misread as improvisation. For Sarasvathy, the approach offers five decision-making principles: start with the means—the entrepreneur’s skills, knowledge, and social relationships; form partnerships to seek business opportunities together and share resources; focus on downside risk and worry more about affordable loss than profits to achieve; take advantage of contingencies rather than thinking of them as obstacles; and control the future, which is uncertain but controllable because entrepreneurs can influence trends, create new markets, and face new challenges. The affordable-loss principle is particularly important: it reframes the decision from “what is the expected return?” to “what is the most I am willing to lose to find out?”—a question that can always be answered, even when probabilities cannot.

Research suggests a key distinction: while some founders adapt deliberately, others rely on instinct and rationalise their decisions only afterward. This makes it hard—even for them—to tell whether they are making strategic choices or simply reacting under pressure.⁷ The discipline effectuation demands is not boldness—it is the willingness to make the logic of your decision explicit before you act, not after.

Three practical instruments

1. The reversibility test

Not all ambiguous decisions carry equal stakes. The most operationally useful sorting mechanism comes from Jeff Bezos’s 2016 letter to Amazon shareholders. In that letter, Bezos encouraged allocating decisions into one of two buckets: reversible “two-way door” decisions, which can be changed if they do not work out; and irreversible “one-way door” decisions, which must be made methodically, carefully, slowly, with great deliberation and consultation.⁸

Most decisions only need about 70% of the information you wish you had. Disagree and commit: people can disagree, but once a decision is made, everyone must commit to it. The corollary is equally important: as organisations get larger, there seems to be a tendency to use the heavyweight Type 1 decision-making process on most decisions, including many Type 2 decisions. The end result of this is slowness, unthoughtful risk aversion, failure to experiment sufficiently, and consequently diminished invention. Early-stage founders rarely suffer from moving too fast on reversible decisions; they suffer from treating every decision as though it were irreversible.

2. The pre-mortem

When a decision is genuinely consequential—a one-way door—the pre-mortem is the most cost-effective debiasing tool available. Developed by psychologist Gary Klein and first detailed in a 2007 Harvard Business Review article, the pre-mortem method counters common cognitive biases such as overconfidence and groupthink by encouraging open dissent during the planning phase, rather than waiting for problems to emerge post-implementation.⁹

The mechanism is prospective hindsight. Before a decision is finalised, you mentally jump forward in time to a point where the decision has already been executed—and has failed. From that imagined future vantage point, you look back at the present and ask: what happened? Gary Klein found it improves the ability to identify reasons for future failure by about 30 percent, because it gives people permission to voice the doubts that optimism and group dynamics usually suppress.¹⁰ Nobel laureate Daniel Kahneman highlighted the pre-mortem in his 2011 book Thinking, Fast and Slow as an effective method to mitigate the planning fallacy and optimism bias, encouraging teams to prospectively identify failure causes rather than overlooking risks.

The protocol is brief. Klein’s protocol runs in 20–30 minutes: announce the project has definitively failed; each participant independently writes every reason the failure happened; round-robin sharing builds a complete list; then the team discusses the list, identifies mitigations, and updates the plan. One structural caution: running it with senior executives present collapses the candour the exercise depends on. Participants self-censor reasons that might implicate leadership decisions, missing the most important risks.

3. Assumption mapping before commitment

Lipshitz and Strauss, in their 1997 naturalistic decision-making analysis, identified how real-world decision-makers cope with uncertainty. Based on a literature review of work describing real-world decision-making under varying forms of uncertainty, Lipshitz and Strauss developed a framework for understanding how decision-makers in organisations cope with uncertainty. In their framework, uncertainty is either acknowledged through preemptive action and planning, reduced through rule- or assumption-based reasoning, or suppressed through ignoring information or guesswork.¹¹ The suppression path—ignoring information or guessing—is the one most founders inadvertently take when moving fast.

The practical translation: before committing to any significant course of action, list the three to five assumptions whose failure would invalidate the entire decision. Then ask which of those assumptions can be tested cheaply before the commitment is made. Amazon’s Working Backwards process institutionalises a version of this. Working Backwards is a systematic way to vet ideas and create new products. Its key tenet is to start by defining the customer experience, then iteratively work backwards from that point until the team achieves clarity of thought around what to build.¹² The PR/FAQ document forces the team to articulate assumptions explicitly—and to confront the questions they cannot yet answer—before a single line of code is written or a dollar of capital is deployed.

The meta-skill: distinguishing the type of uncertainty you face

Not all ambiguity is the same, and treating it as homogeneous is itself a decision error. Based on a literature review of work describing real-world decision-making under varying forms of uncertainty, Lipshitz and Strauss developed a framework in which uncertainty is either acknowledged through preemptive action and planning, reduced through rule- or assumption-based reasoning, or suppressed through ignoring information or guesswork. The first two responses are productive; the third is the default under pressure.

A useful diagnostic: ask whether the uncertainty you face is reducible or irreducible. Reducible uncertainty can be addressed by gathering more information—customer interviews, a pilot, a structured experiment. Irreducible uncertainty cannot be resolved before a commitment; it can only be managed by limiting downside exposure and preserving optionality. Thanks to the work of Knight (1921) and Ellsberg (1961), this degree of uncertainty over vague or unknown probabilities is referred to as ambiguity. Different people have a different “taste” for the lack of accurate information about the probabilities of given outcomes and will respond differently. Our preferences towards ambiguity guide the decisions we make under uncertainty. Everyone has a different tolerance for the level of risk they are comfortable accepting and the amount of uncertainty they are happy to make decisions within.¹³

That tolerance is not fixed. It can be trained—through deliberate exposure to small, reversible decisions made under ambiguity, with fast feedback loops. While novice entrepreneurs tend to rely on predictive techniques, expert entrepreneurs know better than trying to predict the future. Instead, they manage uncertainty through non-predictive control. Non-predictive control is not passivity; it is the active shaping of the environment through commitments, partnerships, and small bets that generate information rather than consume it.

The founder’s edge is not superior information—it is a higher tolerance for acting on incomplete information, combined with the discipline to know which incompleteness is acceptable and which is fatal.

What this means

The founders who build durable companies are not the ones who had better data at the start. They are the ones who built better decision architectures for operating without it—who could distinguish reducible from irreducible uncertainty, who ran pre-mortems before major commitments, who applied effectual logic when causal logic had nothing to grip. The data will eventually arrive. The question is whether the decisions made in its absence were disciplined enough to survive until it does.

At FounderWise, the Business Growth Accelerator (a FounderWise brand) works with founders and operators on exactly this kind of structural thinking—building decision frameworks that hold up under the conditions that actually exist, not the conditions that textbooks assume.