← All answers

Taking action

Which assumption should I test first when I am unsure about a direction?

Test first the assumption that is both important enough to change your next decision and weakly supported by current evidence—provided you can examine it safely with a small, reversible action. An uncertain assumption that would not change what you do is not first. A critical assumption that requires material risk, cost, consent, or an irreversible commitment is not ready for a personal experiment. Use the SHIFT filter to compare significance, evidence strength, inspectability, fallout, and timing without pretending the ranking is mathematically certain.

Write the assumptions as testable statements

Begin with the direction you are considering and complete the sentence “This next step makes sense only if…” several times. An assumption might concern interest, available time, access, cost, skill fit, another person's agreement, or whether the core activity is tolerable enough to repeat. Keep facts, hopes, and constraints separate.

The 2026 Green Book recommends making the evidence, research, and assumptions behind judgments explicit. It also says appraisal should be proportionate to likely cost and risk. That guidance governs public decisions, not personal direction finding, but the interpretation is useful: assumptions that remain vague cannot be compared or tested carefully.

An assumption is not first merely because it feels urgent. It is first when learning whether it holds could change what you do next.

Use the SHIFT filter

SHIFT is a Forever Free Compass framework, not an external research finding. Review each assumption in plain language; do not invent precise probabilities or multiply the fields into a scientific-looking score.

  1. Significance: if this assumption is wrong, would you stop, redesign, delay, or choose a different direction?
  2. History of evidence: what reliable evidence already supports it, and how well does that evidence transfer to your situation?
  3. Inspectability: can a bounded action produce an observable result connected to this assumption?
  4. Fallout: what time, money, exposure, obligation, safety risk, or effect on other people could the test create?
  5. Timing: can the signal arrive before you must make a larger or less reversible commitment?
SHIFT patternPriorityResponse
High significance, weak evidence, inspectable, low fallout, timelyTest firstRun the smallest real-world check and define the evidence rule in advance
High significance, strong transferable evidenceUsually lowerUse the existing evidence unless your context differs in a decision-relevant way
Low significance, weak evidenceLowerDo not spend the first test on something that would not change the next action
High significance, high fallout or poor inspectabilityDo not experiment casuallySeek safer evidence, consent, expert guidance, or a redesigned step

Prioritize criticality and evidential weakness together

The UK government's 2026 Test and Learn guidance sorts assumptions by two dimensions: how critical they are to the intended outcome and how strong the supporting evidence is. It recommends focusing testing on critical assumptions with a weak evidence base instead of testing every assumption equally.

For a personal direction, translate criticality into a decision question: “If this turned out to be false, would my next step change?” Translate evidential weakness into an evidence question: “What supports this belief, and does it apply here?” An assumption can be uncertain without being important, and important without needing a new test.

Match the test to the uncertainty

Government functional standard GovS 010 says sources of uncertainty and their impacts should be identified, and that analysis should be proportionate to the decision being supported. It also warns against unwarranted confidence in one option.

Choose a test that can distinguish plausible outcomes connected to the assumption. If the assumption is “I would use this skill repeatedly,” completing two samples of the core activity may be more informative than reading another general overview. If it is “the schedule can work,” a calendar simulation or one bounded trial may be enough to expose a constraint. Record what different observations would change before you begin.

A useful test changes the next decision. More activity is not the same as more decision-relevant evidence.

Use a what-if check when a live test is unnecessary

The Green Book says low-cost, low-risk proposals may use simple what-if questions about the assumptions behind an appraisal. It also describes flexibility as valuable when uncertainty is significant or an initial investment would be difficult to reverse.

Ask: “If this assumption were false, would the small next step still be acceptable?” If yes, you may not need to test it first. Ask: “What reasonable change would make this step unacceptable?” If a modest change breaks the plan, shrink the commitment or gather evidence before acting. This is a plain-language planning adaptation, not a formal appraisal.

Protect people and preserve a stop condition

Do not use SHIFT to justify testing an assumption through another person without their agreement, bypassing duties or confidentiality, or accepting a downside you cannot repair. A critical assumption is not automatically safe to test.

When health treatment, legal rights, regulated finances, immediate safety, employment duties, or material effects on another person are involved, obtain qualified guidance and required consent. For a low-stakes personal test, set a time and cost cap, define the observable signal, name the stop condition, and schedule a review before a larger commitment.

Review what the test actually changed

After the test, separate the observation from the conclusion. Record what happened, how reliable and relevant the signal is, which assumption changed, and what remains unknown.

Stop when the assumption failed and it was necessary for the direction. Adapt when the direction remains plausible but the method or boundary was wrong. Repeat only when another bounded observation could materially improve the decision. Expand by one step when the evidence supports it and the new commitment remains proportionate.

Original contribution

The Forever Free Compass take

Forever Free Compass uses the SHIFT filter as an original planning framework: Significance, History of evidence, Inspectability, Fallout, and Timing. SHIFT gives priority to an assumption that could change the next decision, has weak or poorly transferable support, can produce an observable signal, keeps downside and effects on others bounded, and can inform the choice before a larger commitment. It is not a validated assessment, probability model, risk score, or rule that the most uncomfortable assumption must come first.

Sources

  1. Test and LearnHM Treasury and Evaluation Task Force. Prioritising assumptions by criticality and evidential strength; focusing tests on critical assumptions with weak evidence; early prototyping and proportionate learning.
  2. Government functional standard GovS 010: AnalysisUK Government Analysis Function. Identifying sources and impacts of uncertainty; proportionate analysis; appropriate assumptions, options, and scenarios; avoiding unwarranted confidence.
  3. The Green Book (2026)HM Treasury. Making evidence and assumptions explicit; proportionality to cost and risk; simple what-if questions for low-cost, low-risk proposals; preserving flexibility under uncertainty.

This is educational reflection, not medical or mental-health care. If distress is persistent, severe, or involves immediate safety, contact a qualified professional or local emergency service.

Use the SHIFT filter on the assumptions behind one possible direction.

Start your free compass