Auditing Your Assumptions Before Drafting a Future Scenario

Why an assumption audit comes before scenario drafting

A scenario can sound convincing while resting on several unsupported beliefs. A team might imagine that AI assistants become standard in public administration, that citizens accept automated decisions and that existing institutions can adapt their procedures quickly. Each statement is a separate assumption. If the team combines them into one narrative without examining their dependencies, the resulting scenario may conceal more uncertainty than it explains.

An assumption audit makes those beliefs explicit before writing begins. It does not require the team to eliminate uncertainty. Instead, it helps participants understand which conditions are observable today, which are interpretations and which are propositions about the future. This distinction improves the quality of later discussion and makes it easier to compare competing scenarios on consistent terms.

The assumption audit worksheet

Begin with a planning question and a defined time horizon. For example: How might AI-assisted public services change citizen access by 2035? Ask participants to write down every statement that must become true for their preferred future to occur. Keep each assumption narrow enough to examine independently. A statement such as AI will transform government is too broad. A more useful version is that local agencies will adopt AI-assisted systems for at least some routine administrative interactions.

For each statement, record its evidence, uncertainty, importance, dependencies and falsification condition. Use a simple three-point scale for importance and uncertainty: low, medium or high. A highly important, highly uncertain assumption deserves attention early because a change in that condition could substantially alter the scenario. Add an owner who can identify relevant evidence as the planning work continues.

• Assumption: the specific proposition the scenario requires.
• Evidence: the present-day information supporting it.
• Uncertainty: how difficult it is to anticipate whether it will hold.
• Importance: how strongly the scenario depends on it.
• Dependency: the other conditions that must also hold.
• Falsifier: an observable development that would weaken the proposition.

Separating facts, interpretations and assumptions

A common source of planning error is treating an interpretation as if it were an established fact. Consider the statement: Organizations are adopting AI quickly. It combines a possible observation with a broad conclusion. A more disciplined record might say that a named organization announced an AI pilot on a particular date. The interpretation might be that the announcement indicates growing institutional interest. The future assumption might be that adoption will become widespread across comparable organizations by 2030.

These three statements require different kinds of evidence. The announcement can be confirmed through a primary source. The interpretation needs comparison with other developments. The future assumption requires a defined adoption threshold and a time horizon. Teams should also record the difference between an announced intention, an operational deployment and a sustained change in everyday practice. Those stages are frequently confused in technology discussions.

Using a futures wheel to expose indirect consequences

The futures wheel is a structured way to explore the consequences of a central change. Place the proposed development in the center of a page. Around it, write first-order consequences: effects that follow relatively directly. Then add second-order consequences that arise from those initial effects. A third ring can be useful when the team needs to examine longer causal chains, but it should not become an exercise in inventing increasingly remote possibilities.

For example, widespread use of AI-assisted public-service applications might produce faster initial responses. A second-order effect could be increased demand for human review when automated decisions are disputed. A third-order consequence might involve new staffing models for appeals and case resolution. Each connection should be treated as a hypothesis. Ask what evidence would support the relationship and whether another factor could produce the same effect.

• Center: the change being explored.
• First ring: direct operational, social or institutional effects.
• Second ring: consequences of the first-order effects.
• Third ring: longer-term effects that merit further examination.

Backcasting method steps for testing the path to a future

Backcasting starts with a clearly described future condition and works backward to identify the developments required to reach it. It is useful after an initial assumption audit because it reveals missing steps between the present and the proposed endpoint. If a team imagines that AI-assisted public services are routine by 2035, it should ask what must be in place by 2032, 2029 and the current planning year.

A practical sequence has five steps. First, define the future condition in observable terms. Second, identify the institutional, technical and social conditions required. Third, work backward to identify intermediate milestones. Fourth, examine which milestones depend on unresolved assumptions. Fifth, identify present-day actions that remain useful even if the preferred future does not materialize. The timeline should represent dependencies, not an invented guarantee that events will occur on schedule.

Prioritizing assumptions and identifying falsifiers

Once the team has mapped its assumptions and causal links, prioritize the ones with both high importance and high uncertainty. For each, define a signal that would strengthen the assumption and another that would weaken it. A useful falsifier must be observable. For example, a team assuming broad public acceptance of automated administrative decisions might monitor the proportion of services retaining human appeal channels, changes in public consultation requirements and formal restrictions on automated decision-making.

Avoid vague conditions such as adoption fails or trust declines. Define a threshold and a time window where possible. A planning team might specify that fewer than three of ten selected pilot organizations move into sustained operational use by the end of a defined review period. The number is a team-selected monitoring threshold, not a universal industry benchmark. Its value lies in making disagreement and subsequent reassessment possible.

From the worksheet to a better planning conversation

An assumption audit should end with a decision about what the team needs to investigate, what it can safely leave uncertain and which assumptions deserve monitoring. Not every uncertainty requires research. Some are peripheral to the decision. Others may be important but impossible to resolve in the available time. Record those limits explicitly so that a polished scenario does not give them false certainty.

A useful workshop can be completed in 60 minutes with four participants: 10 minutes to define the planning question, 15 minutes to list assumptions, 15 minutes to map consequences, 10 minutes to backcast and 10 minutes to identify falsifiers. Before the next session, transfer the five most consequential assumptions into a shared register and assign each one a monitoring owner.

Questions readers ask