A strategy is a structure resting on beliefs, and most organisations can name the structure precisely while being unable to name the beliefs holding it up.
The investment paper runs to forty pages. Thirty-eight of them are analysis: market sizing, competitive position, financial modelling, delivery approach, benefits profile. Two of them, near the back, are a table headed Assumptions.
Approving the paper looks like a straightforward decision. The analysis is competent, the logic holds, the numbers work. The committee approves.
What has actually happened is that the organisation has committed capital to a structure whose load-bearing elements are described on two pages that most attendees skimmed — because the convention in almost every enterprise is that the analysis is the substance and the assumptions are the disclaimer.
The convention is precisely backwards. The analysis is a set of consequences. The assumptions are the premises those consequences depend on. When an investment fails, it is almost never because the arithmetic was wrong. It is because something that was assumed turned out not to be so, and nobody was watching.
Portny's account of defining work under uncertainty places assumptions as one of four components of a statement of work, alongside purpose, objectives and constraints. His definition is worth holding onto: an assumption is something a plan is built upon that nobody yet knows to be true.
Read at enterprise level, that is not a documentation requirement. It is a description of where strategic exposure actually sits.
The Strategic Context
Executives are trained to interrogate conclusions. Boards are constituted to test recommendations. Most governance forums are designed around the question do we agree with this proposal?
Very little of that machinery is pointed at premises. When an assumption is challenged in a committee, the challenge usually takes the form of scepticism about a number — is that growth rate realistic, is that timeline achievable — rather than an examination of the belief structure the number rests on. The distinction matters, because a wrong number produces an error of degree while a wrong premise produces an error of kind.
There is a well-established discipline that inverts the usual question. Rather than asking whether a strategy is likely to succeed, it asks: what would have to be true for this to work? The value of the inversion is that it converts an argument about confidence into a list of testable propositions — and a list of testable propositions can be assigned, monitored and, crucially, disproved.
Most organisations produce something that superficially resembles this list. Almost none use it as an instrument. The assumptions table is written to satisfy a template, approved with the paper, filed with the paper, and never revisited — including at the moment when one of its entries becomes demonstrably false.
What Leaders Commonly Misread
That assumptions are a category of risk. They are not, and conflating them costs precision. A risk is something that may happen. An assumption is something being treated as though it were already settled. The organisation's exposure to a risk is contingent; its exposure to an assumption is immediate and continuous, because the plan is already built on it. Filing assumptions inside the risk register buries the immediate exposure inside a list of contingent ones.
That recording an assumption is the same as managing it. An entry in a table has no owner, no verification method and no consequence. Managing an assumption means someone is accountable for knowing whether it still holds, and something specific happens when it does not.
That constraints are administrative. Portny lists constraints alongside assumptions for good reason: they are restrictions on how the work may be approached. At enterprise level, constraints are where strategy meets the organisation's actual boundaries — regulatory limits, capital availability, sovereign requirements, industrial agreements, existing commitments. Treating them as background rather than as design parameters produces strategies that are elegant and unbuildable.
That the assumptions most likely to fail are the ones written down. They are usually not. The written assumptions are the ones someone noticed. The dangerous ones are typically so widely shared inside the organisation that no one thought to state them — that a partner will remain a partner, that a capability the enterprise has always had will still be there, that a regulatory posture will hold.
Reframing the Issue
The reframe: the assumptions register is the strategy's structural drawing, and it should be read the way an engineer reads one — looking for what is carrying load.
Not every assumption matters equally. A plan may rest on forty stated beliefs, of which perhaps four are load-bearing: if they fail, the strategy does not degrade, it collapses. The remaining thirty-six affect efficiency, cost or timing.
The executive discipline is not to track all forty. It is to identify the four, and to build genuine organisational attention around them. That is a substantially different activity from maintaining a register, and it produces a substantially shorter document.
A defence acquisition illustrates the distinction; the example that follows is hypothetical. Such a program rests on assumptions about threat environment, sovereign industrial capability, technology maturity, workforce availability, sustainment costs over decades, and interoperability with allied systems. All are recorded. But only some are load-bearing — and which ones they are is a question of program design, not of documentation. An assumption about sustainment cost may be recoverable through budget adjustment. An assumption that a specific sovereign capability will exist by a specific date may not be recoverable at all, because if it fails, the option set does not narrow — it disappears.
The load-bearing test is therefore not how likely is this to be wrong? but what remains possible if it is?
Distinguishing the Four Elements
Portny's four components map cleanly onto four executive questions, and separating them prevents a common failure in which all four collapse into narrative.
Purpose — why the work was established and by whom, what scope it covers, and the broad strategy for achieving it. The test: could a reader tell what the organisation is trying to change, and on whose authority?
Objectives — the specific results to be achieved. The test: could two independent executives agree whether they had occurred?
Constraints — the boundaries within which any solution must sit. The test: are these genuine limits, or preferences that have acquired the language of limits? This distinction is worth forcing, because organisations routinely accept as constraints things that are actually unexamined defaults, and thereby exclude the best available options without ever considering them.
Assumptions — what is being treated as settled but is not known. The test: which of these, if false, would make the objectives unreachable rather than merely harder?
The discipline is in the separation. A proposal that cannot distinguish its constraints from its assumptions has not been thought through, whatever the quality of the analysis built on top.
Decision Framework
For any material investment, apply this before approval.
| Step | Question | Output |
|---|---|---|
| 1. Surface | What must be true for this to succeed? | Written list, premises only — not risks |
| 2. Separate | Which are genuinely unknown, and which are constraints? | Two lists, not one |
| 3. Weight | If this fails, does the strategy degrade or collapse? | Load-bearing set, typically three to six |
| 4. Assign | Who is accountable for knowing whether each still holds? | A name per load-bearing assumption |
| 5. Instrument | What observation would show it has failed? | A specific, cheap, observable test |
| 6. Pre-decide | What happens if it fails? | Response agreed before commitment |
| 7. Review | When is each re-examined? | A date, tied to a governance forum |
Step 6 carries most of the value and is most often skipped. Deciding the response to a failed assumption before capital is committed is a different conversation from deciding it afterwards — because before commitment, stopping is still an available answer.
From Strategy to Execution
Immediate. For each of the enterprise's three largest active commitments, ask one question: name the three beliefs this depends on, and tell me who is watching them. The quality of the answer, and the speed of it, is a direct reading of governance maturity.
Medium term. Move the load-bearing assumptions out of the investment paper and into the standing governance record. An assumption filed with an approval document is inert. An assumption on a board or portfolio agenda, with an owner and a review date, is an instrument. [Related article: Drivers, Supporters and Observers: Who Is Allowed to Change What Your Program Is For]
Long term. Build the habit of writing down what would disprove a position. This is uncomfortable in organisations where confidence is a proxy for competence, and it is the single most reliable protection against the failure mode where an enterprise defends a strategy long after its premises have gone.
Signals to Monitor
- Assumption tables that do not change between reporting periods. Evidence of a document, not an instrument.
- Constraints that soften when challenged. Suggests preferences were recorded as limits, and the real option set was never examined.
- Load-bearing assumptions with no named owner. The exposure exists regardless; only the accountability is missing.
- Approvals citing analysis depth as confidence. Thirty-eight pages of consequence do not strengthen two pages of premise.
- Assumptions discovered rather than declared. When a failed assumption surfaces through an incident rather than through review, the register was not operating.
Questions for the Leadership Team
- What are the three beliefs our largest current investment depends on, and which of us is accountable for each?
- Which of our stated constraints are genuine limits, and which are defaults we have never tested?
- For our most significant strategic commitment, what observation would tell us the premise has failed — and would we see it?
- Have we agreed, in advance, what we would do if a load-bearing assumption proved false?
- What does this organisation assume so universally that nobody has written it down?
Closing Perspective
Enterprises rarely fail because they reasoned badly from their premises. They fail because the premises changed and the reasoning continued.
The assumptions register is the only governance artefact designed to catch that, and in most organisations it has been demoted to an appendix — recorded once, to demonstrate diligence, and then left behind by the very analysis it holds up. Promoting it is not a process improvement. It is the difference between a strategy the organisation is governing and one it is merely executing.
Related article: Uncertainty Is the Case for Planning, Not the Excuse Against It
Related article: Drivers, Supporters and Observers: Who Is Allowed to Change What Your Program Is For
Related article: Why "The Principles Apply to Any Project" Is Only Half True
About the author
Kevin Jogin is Founder & Principal Advisor at EraNorth. Meet the Founder.