Deep uncertainty

Deep uncertainty exists when an organisation cannot confidently define all relevant future states, assign probabilities to them or fully understand the consequences of its own actions in advance. It is the condition in which planning alone is not enough because the situation will reveal part of itself only through action.

In brief

Deep uncertainty is stronger than ordinary uncertainty. It is not only that the company lacks a number. It is that the structure of the problem is partly open.

The organisation may not know which variables will matter, how stakeholders will react, how internal routines will adapt, or which second-order effects will appear.

In this condition, the useful question is not “what is the exact forecast?” but “how do we act without closing too many options before learning enough?”

Operational definition

A decision involves deep uncertainty when at least one of the following is true.

The possible future states cannot be fully listed.

The probabilities are not reliable enough to guide the decision.

The consequences depend on interactions that are not yet observable.

The organisation itself will change as a result of the decision, making pre-action analysis incomplete.

This is common in organisational change, acquisition integration, first-time managerial hiring, new technology adoption and market repositioning.

Why it matters for SMEs

SMEs often operate with thin data and strong founder intuition. That can be an advantage, but it can also hide deep uncertainty behind confidence.

A first ERP implementation, a move into a new market, a new management layer, or a post-acquisition integration are not only risky. They are uncertain because the organisation will learn its constraints while acting.

If the company treats deep uncertainty as a normal planning problem, it may overcommit early and discover too late that the assumptions were wrong.

Observable signals

Look for disagreement about what the decision is really about.

Look for experts who cannot agree on the relevant variables.

Look for plans that require the organisation to behave in ways it has never tested.

Look for decisions whose consequences depend on adoption, trust or informal routines.

Look for large commitments before learning milestones.

Common mistakes

The first mistake is demanding a forecast where the structure is still unknown.

The second mistake is using analysis to delay every action. Some knowledge appears only after a controlled step.

The third mistake is choosing an irreversible path too early.

The fourth mistake is ignoring organisational capacity. Under deep uncertainty, the company must learn and execute at the same time. That requires cognitive margin.

Operational example

A founder hires the first operations director to “structure the company”. The job description looks clear, but the real uncertainty is deep: how much authority the founder will actually release, how teams will respond, which decisions will remain informal, and whether the role will become a true operating layer or a symbolic title.

A risk plan might define targets and review dates. A deep-uncertainty approach adds a 90-day role-gap review, explicit decision rights, protected decision boundaries and a stop rule for unresolved authority conflicts.

The company is not pretending to know everything in advance. It designs learning into the first cycle.

Diagnostic questions

Which parts of the situation cannot be known before action?

Which assumptions will become visible only after people start behaving differently?

Which commitment would be difficult to reverse?

What is the smallest step that creates useful learning?

What would make the company stop, pause or redesign the initiative?

Does the organisation have enough margin to learn while executing?

Practical implications

Under deep uncertainty, use staged commitments. Define learning questions. Preserve options. Avoid big-bang decisions when a pilot can expose the structure.

Build review gates into the work. Use stop rules. Keep some slack. Do not measure success only by adherence to the original plan; measure how quickly the organisation learns and adjusts.

MARTRO reading

In MARTRO’s reading, deep uncertainty is a condition to be designed around, not an embarrassment to remove rhetorically. It is where diagnostic sequencing matters most.

The method therefore asks which intervention should come first, which options must remain open, and which decision would close doors before evidence exists.

Frequently asked questions

Is deep uncertainty just high risk? No. High risk can still have known variables and probabilities. Deep uncertainty means the structure of the future state is partly unknown.

Can analysis solve deep uncertainty? It can reduce some ignorance, but it cannot remove what will only become visible through action.

Should we avoid decisions under deep uncertainty? No. Avoiding all action can also close options. The point is to act in staged, reversible and learning-oriented ways.

What tools help most? Pilots, staged commitments, optionality, stop rules, role-gap checks and clear review gates.

Why is this important for SMEs? Because many SME transformations fail by acting as if the organisational consequences were already knowable when they were not.