Assumptions & Constraints

Every Decision rests on conditions that are not fully under management's control.

Some are things we are treating as sufficiently true. Others define what we are allowed or able to do.

AIMR separates these as assumptions and constraints.

An assumption is a condition treated as sufficiently true for the Decision Process even though it has not been fully established or verified.

A constraint is a condition that limits the available choices, authorization, or legitimate downstream action.

The distinction matters because the management response is different.

An assumption asks:

What are we relying on?

A constraint asks:

What limits us?

Assumptions carry the argument

Management rarely knows everything needed for a Decision with certainty.

We assume a supplier will deliver, demand will remain within a range, funding will remain available, a policy interpretation will hold, or a technical capability will perform as expected.

That is not necessarily weak management. Decisions often require action before every uncertainty can be removed.

The important discipline is to make the material assumptions visible enough to govern.

Which assumptions materially support the Decision?

How strong is the evidence behind them?

What uncertainty remains?

What would happen if an assumption proved false?

Which changes would be significant enough to require further management attention?

An assumption does not become a fact merely because it appears in a polished business case or AI-generated analysis.

Constraints define the legitimate space

Constraints work differently.

A regulatory requirement may rule out an option. A contractual commitment may limit timing. A Decision Mandate may restrict what a manager can authorize. A capacity limit may narrow the practical choices. An enterprise policy may require additional governance.

Some constraints can be resolved. Others remain and define the permissible Decision space.

A dependency is not automatically a constraint. Incomplete information is not automatically a constraint. Uncertainty is not automatically a constraint.

They become constraining when they materially limit what the Decision Process can legitimately or practically do.

Changes do not produce automatic conclusions

An assumption can fail without automatically invalidating the Decision.

A constraint can tighten without automatically requiring escalation.

The effect depends on the particular Decision Context.

A failed assumption might be immaterial. It might be manageable within existing discretion. Or it might materially affect the basis on which the Decision remains appropriate.

Likewise, a constraint might be resolved locally, addressed through delegation, or require a different Decision Jurisdiction.

The important point is not to turn assumptions and constraints into automatic workflow triggers detached from management judgment.

Assumptions should have a management consequence

Making assumptions visible does not mean building an exhaustive assumption register for every Decision. The useful assumptions are the ones that carry enough of the Decision logic that their failure could change management's judgment.

For those assumptions, management should know what evidence supports them, how uncertainty is being treated, and whether any monitoring is warranted. Some assumptions can simply be accepted. Others may justify a Threshold, a review condition, or additional Management Intelligence.

The discipline is proportional. The organization is not trying to eliminate assumption. It is trying to prevent a material assumption from becoming invisible merely because everyone happened to agree with it at authorization.

AI can surface them, not govern them by itself

AI can help identify hidden assumptions, compare them with new evidence, detect constraint changes, and maintain traceability across a large Decision portfolio.

But AI does not independently determine which assumption is acceptable, which uncertainty the organization should tolerate, or which constraint changes the legitimate Decision space unless that judgment has been explicitly governed within its Mandate.

This also makes assumptions useful for later learning. When the outcome differs from expectation, management can ask whether the assumption was unreasonable, merely uncertain, or reasonable but overtaken by events.

What management should look for

For a material Decision, ask:

• What are we treating as true even though it is not fully established?

• Which of those assumptions materially support the Decision?

• What conditions limit our available or legitimate choices?

• Which constraints can be resolved, and which must simply be respected?

• Do we know which changes would materially affect the Decision?

• Are assumptions being confused with facts, and dependencies with constraints?

Boundary

Assumptions are not facts.

Constraints are not simply difficulties.

Dependencies, uncertainty, and incomplete information are not automatically constraints.

A changed assumption does not automatically invalidate a Decision.

A changed constraint does not automatically determine the response.

The management implication

Good Decision governance does not require eliminating uncertainty. It requires knowing where management is relying on conditions that may prove wrong and where legitimate choice is genuinely bounded.

Know what you are relying on. Know what limits you. Know which changes would be significant enough to require the Decision to be reconsidered through its governing arrangements.