Evidence-Based Discovery Planning

Hello. Last lesson ranked your venture assumptions by how much would change if they proved false and how little relevant evidence you have. A high-priority assumption is not yet a research plan: you still need to decide what to learn, from whom, by which method, and what you will do with the findings.

This lesson turns that shortlist into a first discovery round for the heavy-flow opportunity. The aim is to learn whether a specific unmet need persists, not to validate a tech solution by asking whether people like one. By the end, you will have a plan with research questions, evidence to look for, and decision thresholds set before you hear the answers.

Turn assumptions into questions you can investigate

Return to the illustrative top-ranked claim from last lesson:

Among people who describe heavy flow and work shifts with restricted breaks, existing preparations still leave consequential disruption on high-flow workdays.

The corresponding research question is not “Is heavy flow a problem?” That is too broad to guide a decision. Nor is it “Would you use an app to manage it?” That introduces a solution the claim does not require. A more useful question is: During recent high-flow shifts with limited breaks, what happened despite the preparations people made, and what consequences did it have?

This change of wording does three things. It specifies the people and context you need to hear from. It identifies observable evidence—accounts of particular incidents, preparations, and consequences. And it leaves room for the answer “my preparations worked.” That answer could challenge the proposed segment or need; it is not a failed interview.

The GOV.UK Service Manual’s Plan user research for your service offers a concise framework for moving from assumptions to research questions, user groups, methods, and focused rounds. Read it as planning guidance, not as a prescribed schedule for a startup.

For this first round, distinguish two questions that are easy to collapse:

1. Is there unresolved disruption in this proposed context? Ask about recent high-flow shifts, existing preparations, what occurred, and the effect on work or daily life. This investigates assumption A from last lesson.
2. When disruption occurs, do people seek or want help beyond what they currently do—and what prevents them? Ask what they have tried, looked into, avoided, or been unable to access, and why. This investigates assumption B.

Someone might experience substantial disruption yet have stopped looking for help because available options feel inaccessible, ineffective, or uncomfortable. Conversely, someone might try many products but now feel adequately supported. An action is evidence, but its meaning requires context. Do not equate “no additional purchase” with “no unmet need,” or an account of difficulty with proof that someone will adopt your eventual product.

Keep your terms equally precise. In this round, eligibility can rely on people describing their own flow as heavy or difficult to manage; an interview cannot establish a clinical diagnosis of heavy menstrual bleeding. Document participants’ wording and experiences without treating self-description as a clinical measure.

Choose methods for the claim—not for the product you imagine

A method is useful when the evidence it produces bears directly on the question. For A and B, semi-structured interviews about recent experiences are a sensible first method. They let you investigate sequence and context: what a person anticipated, what they brought or arranged, what actually happened, and which consequences mattered to them. They also let an unexpected need surface. A short review of credible secondary evidence can help you check whether themes are plausible beyond your interview pool, but it cannot establish the pattern in your particular proposed segment.

Watch this excerpt from Customer Discovery: What Do You Ask, presented by Techstars with Justin Wilcox. It demonstrates why accounts of past events and existing workarounds are more useful at this stage than a pitch followed by predictions about future use.

The interview is not the method for every assumption on your map. If you later choose a private-record concept, a participant can explain concerns about sharing information, but actual usability and trust in that specific interaction will need a carefully designed concept or prototype test. Stated willingness to pay is not payment behavior. And customer interviews cannot determine whether a health-related feature or claim meets regulatory requirements. Match each later test to its claim rather than asking one set of conversations to “validate the startup.”

A first round might involve eight eligible adults who describe flow as heavy or difficult to manage and have recently worked during flow with limited opportunities for breaks. Seek variety within that context—for example, in how predictable breaks are, whether shifts can be changed, and how adequately current preparations work. Eight is a practical qualitative starting point, not a representative sample or a rule that eight interviews prove the market exists. If a pattern is unclear, adjust recruitment and conduct another round.

Because the subject is intimate, the plan should specify what you will not do. Do not request diagnoses, images, cycle logs, or other detailed health records simply to investigate a workday-management question. Participation and note-taking require an appropriate consent process; the next module develops that process and the interview guide. If someone raises a health concern, do not interpret their symptoms or offer a diagnosis. For now, plan to record only the details needed to understand the experience and decision.

Set decision thresholds before seeing the findings

A research question tells you what you want to learn. A decision threshold tells you how findings would change your next move. Without one, a vivid story supporting your idea can outweigh several quieter accounts that contradict it.

Strategyzer’s Validate Your Ideas with the Test Card provides a compact way to make the hypothesis, test, measure, and success threshold explicit.

For qualitative discovery, a threshold is a provisional decision rule, not a statistical cutoff. First define what qualifies as evidence. In the example below, a concrete account identifies a particular recent shift, what the participant prepared or did, and what happened. Consequential disruption might be an unplanned interruption, a leak with a practical consequence, a missed task, or worry that prompted a meaningful change in routine. The participant’s account—not your assumption about what should be difficult—determines the consequence.

Here is a workable first-round plan. Its numbers are illustrative choices to be revised for your actual recruitment, not estimates of prevalence.

These rules give you a reason to continue, revise, or investigate further; they do not authorize building a product. Nor does meeting A’s threshold establish B: people can have a difficult experience without wanting the particular additional help you envision. If A is substantially contradicted, reconsider whether testing B in this same segment is still the best use of the next round.

Numbers alone can conceal weak evidence. Eight people recruited through one closely connected group may share circumstances that do not transfer to others with restricted breaks. Several mentions of “worry” may describe very different effects. Record the nature and source of each account, any contradictory cases, and gaps in whom you reached. Your qualitative-research experience is valuable here: retain enough context to explain why a pattern appears rather than turning eight accounts into a prevalence claim.

Make the plan operational and revisable

A usable discovery plan fits on one page. It names the ranked assumption and research question, eligibility and variation sought, method, evidence to capture, decision rule, and next action. It also states what the method cannot establish.

For this round, your one-page plan could read:

Decision at stake: Whether to keep restricted-break work shifts as a promising first context for further discovery.
Participants: Eight adults with recent experience of self-described heavy or difficult-to-manage flow during shifts with limited breaks; seek varied circumstances and current levels of success with preparations.
Method: Consented, semi-structured interviews focused on specific past shifts and current workarounds, with minimal necessary notes. Review relevant secondary evidence separately.
Evidence: Concrete incidents, preparations, consequences, attempts to improve the situation, barriers, and counterexamples. Keep participant descriptions distinct from your interpretations.
Decision: Apply the provisional A and B thresholds above. Document where the accounts diverge before choosing whether to continue with this context, recruit a different group, or revise the opportunity thesis.
Not established: Clinical status, frequency of these experiences in the wider population, willingness to use a recording tool, willingness to pay, or safety and regulatory viability.

After the round, update the assumption map from last lesson. Move a claim toward “more evidence” only to the extent that the findings address its specific wording. If interviews reveal a more consequential problem in a different context, revise the thesis and write a new question rather than stretching the old threshold to fit it. Thresholds help protect a decision from wishful thinking; they must not prevent you from noticing an unexpected result.

Takeaways

An evidence-based discovery plan connects a consequential assumption to a focused question, a method capable of answering it, and a precommitted decision rule. For the proposed heavy-flow opportunity, recent accounts of preparations, disruptions, and workarounds can test whether a particular context deserves further attention. They cannot diagnose participants, measure market-wide prevalence, or validate an app.

Next, you will make this first round ready to conduct responsibly by defining inclusive participant criteria and a recruitment screener for varied heavy-flow experiences.