Taking action
How do I turn an assumption into a testable prediction?
Turn an assumption into a testable prediction by naming the decision it affects, choosing one uncertain condition that must hold, stating one action and context, defining what you could observe by a set time, and deciding in advance how the result would change your next step. A useful sentence is: “If I do X in context Y, then I expect Z by time T, because R.” The prediction should reduce uncertainty, not promise the final outcome.
Start with the decision, not the activity
A testable prediction connects one uncertain belief to something you can do and observe. Start by writing the next decision the evidence could inform. Then choose one assumption that matters to that decision, state the smallest appropriate action, name the context, and describe an observable cue and time. If the result could not change what you do next, the activity may create information without answering the decision.
The UK Cabinet Office playbook describes testing as an assumption and related hypothesis, an activity, and a way to observe and record results. It also says to set decision criteria early, focus on assumptions that combine high impact with low certainty, and match the amount of evidence to the size of the decision. That is formal public-service guidance; the proportionate transfer here is a short prediction card for ordinary, low-stakes reflection.
Write a PULSE prediction card
PULSE is a Forever Free Compass framework, not an external research finding. Write one sentence or bullet for each field before acting.
- Purpose: name the literal choice the evidence could inform, such as whether to repeat, adapt, pause, or stop a low-stakes next step.
- Uncertain link: state one condition that must hold for the action to help. Prefer an assumption that matters to the desired outcome and currently has weak or incomplete evidence.
- Light action and context: specify the smallest appropriate thing you will do, where or under what conditions, and what you will keep stable enough to interpret the observation.
- Signal and time: name the observable behavior, event, feedback, or record you expect, plus when you will look. Avoid words such as better, successful, or fit unless you define what would be visible.
- Evidence rule: decide before the test what result would support repeating, adapting, pausing, or stopping. Record mixed or negative evidence as learning rather than rewriting the prediction afterward.
Use if–then–because, then add the decision rule
The Cabinet Office playbook offers a hypothesis pattern: if an action is taken, then a predicted change should occur, because of a reason grounded in a theory of change or existing evidence. PULSE keeps that structure and adds context, time, and explicit decision use.
Example: “If I attend one public beginner session this week, then I will be able to name one task I want to try again and one practical barrier by Friday, because direct exposure should reveal more about fit than reading alone. If I can name both without crossing a hard boundary, I will plan one second session; otherwise I will revise the option or the assumption.” The observation does not prove long-term fit or predict a final outcome.
| Vague assumption | Testable prediction | Interpretation limit |
|---|---|---|
| I would probably enjoy this direction | If I try one representative task in the real setting, I expect to identify one part I would willingly repeat by the review date | One willing repeat is an early signal, not proof of durable satisfaction |
| This routine will fit my week | If I schedule the smallest realistic version twice, I expect to complete it without displacing a named non-negotiable | A short trial does not establish long-term sustainability |
| People affected will find the idea useful | If I show a bounded draft to people who may use it, I expect feedback on one useful element and one obstacle | Feedback from a small, consented group does not represent everyone |
Keep the assumption and outcome distinct
Government theory-of-change guidance defines assumptions as underlying conditions needed for intended change and recommends making them explicit, checking the strength of supporting evidence, and planning how they will be monitored and reviewed. A testable prediction should therefore examine one link in the path, not silently replace the desired outcome with an easy-to-observe activity.
The 2026 Test and Learn guidance prioritizes critical assumptions with weak evidence and separates early, proportionate testing from later formal evaluation. Early observations can inform design, feasibility, or the next test; they should not be promoted into causal proof, general conclusions, or a demonstrated final outcome.
Keep the test proportionate and safe
For a low-stakes, reversible choice, one PULSE card and one review point may be enough. Make the action smaller if the cue cannot be observed without excessive time, cost, intrusion, or burden. Stop if the test would cross consent, privacy, safety, legal, financial, employment, housing, or other non-negotiable boundaries. Do not treat safeguards as variables to test.
For consequential decisions or claims about effects on other people, use the relevant formal process and qualified guidance. PULSE does not choose a research design, establish a sample size, validate a measure, demonstrate causation, or make a risky action appropriate. Its narrow purpose is to turn one low-stakes assumption into a prediction that can inform one next decision.
Original contribution
The Forever Free Compass take
Forever Free Compass uses PULSE as an original low-stakes prediction card: Purpose, Uncertain link, Light action and context, Signal and time, and Evidence rule. PULSE adds an explicit decision-use field to a simple if–then–because prediction so a test is bounded by what the reader needs to learn next. It is not a validated experimental, causal, clinical, safety, legal, financial, employment, housing, or professional method and does not establish that an action will produce an outcome.
Sources
- Test, Learn and Grow Playbook — Testing, learning and iterating — UK Cabinet Office. Assumption-hypothesis structure; activity and observation; decision criteria; high-impact, low-certainty assumptions; if–then–because pattern; proportionate evidence; next-decision use.
- Test and Learn (HTML) — HM Treasury and UK Government Evaluation Task Force. Critical assumptions; evidential strength; proportionate methods; early testing; mechanisms; context; separation of exploratory learning from later causal evaluation.
- The Theory of Change Process – Guidance for Outcome Delivery Plans — UK Government Analysis Function. Underlying assumptions; links from activities to outcomes; evidence strength; monitoring and review; risks; dependencies; proportionality.
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.