Taking action
How can I choose a test method that matches what I need to learn?
Choose a test method by writing the decision first, then naming the evidence shape that could inform it. Use interviews or observation to understand experience and context; structured measures or existing records to notice patterns; a prototype or user test to expose practical friction; and a properly designed comparison only when the decision truly needs evidence about difference or effect. Choose the narrowest credible method, set safeguards and interpretation limits, and add a second method only when one cannot answer the learning question.
Start with the decision, not a favorite tool
A test method fits when it can produce evidence that is relevant to one stated learning question and strong enough for the next decision. Write the decision first: repeat, adapt, pause, stop, or investigate further. Then name what you need to learn and what kind of observation could change that choice. Only after those fields are clear should you choose an interview, observation, survey, record review, prototype test, comparison, or another method.
The UK Cabinet Office Test, Learn and Grow playbook says decisions should guide the testing method, evidence, and people involved, and that the planned evidence should match the size of the decision. It explicitly says to match the method to what you need to learn. For ordinary low-stakes reflection, the proportionate transfer is a method-fit card—not an imitation of formal public-service research.
Write a MODE method-fit card
MODE is a Forever Free Compass framework, not an external research finding. Complete one sentence or bullet for each field before collecting new information.
- Move: state the literal next choice and when it must be made. Narrow the question until a plausible result could change that choice.
- Observation: name the evidence shape that would be useful—an explanation, a behavior in context, a repeated pattern, a practical failure point, a difference between versions, or credible evidence of an effect.
- Design: choose the lightest method capable of producing that evidence at the needed credibility. Check existing records before asking people for new data, and add a second method only when the decision needs both breadth and explanation.
- Ethics and limits: define permission, privacy, burden, cost, time, stopping conditions, and what the result will not establish. Escalate when the stakes or causal claim exceed a bounded personal test.
Match the evidence shape to the method
Methods answer different kinds of questions. CDC guidance distinguishes quantitative methods, which can support standardized comparison and trends, from qualitative methods, which can provide deeper explanation and context. It also notes that method choice depends on practical resources, ethics, scope, validity, and reliability. The table below is a low-stakes orientation, not a universal method prescription.
| What you need to learn | Possible starting method | Interpretation boundary |
|---|---|---|
| Why something feels confusing, difficult, or useful | A short interview, think-aloud session, or open reflection | Rich explanation from a small set does not show how common the pattern is |
| What people actually do in a real setting | Direct observation or a bounded user test with permission | Observed behavior may depend on the setting and does not reveal every motive |
| Whether a simple pattern or change appears over time | A short structured measure, existing record, or repeated count | A pattern alone does not explain why it occurred or prove the action caused it |
| Where a draft, routine, or prototype breaks in practice | A walkthrough, prototype test, or small live trial | Early friction can guide revision but does not demonstrate the final outcome |
| Whether one version differs from another | A planned comparison such as an A/B test, with suitable expertise and conditions | A weak or biased comparison cannot support a causal conclusion |
Choose the narrowest credible method
The 2026 Test and Learn annex says early methods should be proportionate to the question and stage, while stronger formal designs belong where the decision and claim require them. It also says using more methods is not inherently better and can add cost without improving insight. Start with one method when it can answer the question within the decision's stakes.
Combine methods only for a stated gap. A small structured measure might show a repeated pattern while one or two consented conversations explore its context. Observation might reveal a practical obstacle while a short follow-up question checks the reason. Keep the findings separate enough to preserve what each method can and cannot say; two weak methods do not become strong evidence merely because they agree.
Protect people and keep conclusions inside the method
CDC guidance says selected methods should be relevant, sensitive, and not harmful, and recommends balancing the amount of data with burden on respondents. Before collecting anything, check whether existing information can answer the question, request only what is necessary, define how it will be handled, and make participation voluntary where other people are involved.
Use MODE only for ordinary, low-stakes learning. Do not use a personal test to bypass consent, privacy, safety, employment, housing, legal, financial, clinical, research, safeguarding, or professional requirements. If the next decision affects other people's welfare, involves material loss, or requires a causal claim, use the relevant formal process and qualified expertise rather than shrinking the method to fit the available time.
Original contribution
The Forever Free Compass take
Forever Free Compass uses MODE as an original low-stakes method-fit card: Move, Observation, Design, and Ethics and limits. MODE makes the required form of evidence explicit before naming a familiar tool, and it uses a one-method default unless the next decision genuinely needs both a pattern and its context. It is not a validated research design, evaluation method, causal-inference procedure, measurement instrument, privacy review, risk assessment, or substitute for consent, formal oversight, required processes, or qualified guidance.
Sources
- Test, Learn and Grow Playbook — Testing, learning and iterating — UK Cabinet Office. Decision-led method choice; evidence proportional to the decision; learning aims; method examples; risks and ethical considerations; interpretation across testing stages.
- Test and Learn (HTML) — HM Treasury and UK Government Evaluation Task Force. Proportionate methods; exploratory and later causal questions; qualitative and quantitative combinations; method selection by assumption, decision, time, and resources; limits of adding methods.
- Step 4 — Gather Credible Evidence — US Centers for Disease Control and Prevention. Question-aligned evidence; qualitative and quantitative method trade-offs; method-selection factors; existing data; mixed methods; relevance, burden, ethics, quantity, and quality.
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.