Turn a technical idea into a testable problem hypothesis and learn what people actually do.
Your learning outcomes
Distinguish an invention from a customer problem
Map users, buyers and other stakeholders
Conduct neutral discovery and maintain an evidence log
CLASS 1.1 / 60 MINUTES
From invention to a useful problem
Separate capability from value
A technical capability describes what an invention can do. A commercial problem describes a costly, risky or frustrating situation experienced by a particular person. Begin by writing both statements separately. SignalBox might measure vibration, but a workshop manager may care about unexpected stoppages or unnecessary maintenance visits. Neither benefit is established until investigated. Avoid attaching a monetary saving to the invention simply because a spreadsheet makes the opportunity look attractive. The first job is to identify whose situation could improve and why that person would act.
Map the system around the user
The person touching the product is not always the person choosing, paying for or approving it. Draw the proposed workflow and identify the operator, maintenance lead, budget holder, IT reviewer, safety specialist and procurement contact. Ask what each participant needs to trust. A useful product can still fail commercially if installation interrupts production or approval requires evidence that the team cannot provide. Record incentives and constraints without assuming every stakeholder wants the same outcome or has the authority to buy.
Write a falsifiable starting hypothesis
Use a sentence that names a segment, a recurring situation, its consequences and the current response. For example: small workshops with older rotating machines may struggle to distinguish routine vibration from a fault worth investigating. This is a teaching hypothesis, not a finding. State what observation would weaken it, such as managers already resolving the issue cheaply through an existing service. A hypothesis that survives every possible answer cannot guide a decision. Keep the proposed solution out of the first round of problem exploration.
PUT IT TO WORK
Your activity
Write three problem hypotheses for SignalBox or your own idea. For each, name the stakeholder, current workaround and an observation that would contradict it.
Produce this
A one-page problem and stakeholder map with three explicit uncertainties.
CLASS 1.2 / 60 MINUTES
Discovery interviews that reduce uncertainty
Ask about recent behaviour
Prepare questions about a specific recent event rather than inviting opinions about your invention. Ask what happened, who became involved, how the issue was resolved and what evidence or spending followed. A compliment about a hypothetical product is weaker than a described purchasing process. Keep the interview conversational and follow unexpected detail. Seek permission before recording, minimise personal information and use neutral notes. The objective is to understand a real workflow, including the parts that make your preferred solution less useful.
Recruit beyond the friendly circle
Friends, colleagues and innovation enthusiasts can help rehearse questions, but they may not represent the intended market. Build a recruitment list across relevant roles, organisation sizes and adoption attitudes. Include people who chose an alternative or decided the problem was not worth solving. Explain that the conversation is research, not a sales commitment, and respect refusals. Keep any incentive modest and transparent. Track who was missing from the sample so that a small set of accessible conversations does not quietly become a claim about an entire industry.
Separate quotation, interpretation and decision
Use three columns in interview notes: what the participant said or demonstrated, what you think it means and what you plan to investigate next. Do not turn a participant's estimate into a verified financial fact. Preserve contradictions instead of averaging them away. In the fictional case, one workshop may want alerts while another wants fewer notifications. That difference might reveal distinct segments, poor framing or a mistaken assumption. The next interview should test the explanation rather than simply collect another supportive quotation.
PUT IT TO WORK
Your activity
Draft a six-question interview guide. Rehearse it with someone acting as an uninterested buyer, then remove leading questions and record what changed.
Produce this
A neutral interview guide and a sample evidence-log entry clearly labelled as a rehearsal.
CLASS 1.3 / 60 MINUTES
Evidence strength and the first decision gate
Rank evidence by the decision it supports
Evidence is useful when it reduces a specific uncertainty. An interview can reveal a workflow, but it does not prove that a sensor achieves an engineering specification. A technical test may show accuracy under controlled conditions without proving willingness to buy. Classify each item by question, method, date, sample, limitation and relevance. Treat an actual paid pilot differently from an expression of interest, while recognising that even a paid pilot may depend on unusual support that cannot be repeated at scale.
Look for patterns without inventing certainty
After several conversations, group observations by situation and role. Count distinct organisations rather than treating three colleagues as three independent buyers. Note both repeated pain and cases where the current workaround is good enough. Avoid claiming statistical representativeness from a convenience sample. The practical question is whether there is enough consistent evidence to justify a next experiment. If the evidence is mixed, narrow the segment or test a different explanation rather than expanding the product until every comment appears satisfied.
Choose a bounded next step
A decision gate connects evidence to a reversible action. Define what must be learned before spending more time or money, what evidence would permit progress and what would trigger a change or stop. SignalBox could next test whether a maintenance lead can use a simple vibration report to make a better investigation decision. It need not build an entire automated platform first. Assign an owner and a review date. Record the decision even when it is to pause, so future enthusiasm does not erase an unresolved weakness.
PUT IT TO WORK
Your activity
Review five fictional or real evidence items. Rank their usefulness for the next decision and write a proceed, change or stop recommendation.
Produce this
An evidence register and a decision memo with a specific next experiment.
MODULE ASSIGNMENT / SELF-REVIEW
Build the evidence.
Produce a discovery pack containing a problem hypothesis, stakeholder map, interview guide and evidence register. Use interviews only with permission; if none are available, clearly label simulated notes and propose a real recruitment plan.
Assignment self-review rubric
Criterion
Points
Problem and stakeholder specificity
25
Neutral questions and ethical recruitment
25
Separation of observation and interpretation
25
Decision rule and next experiment
25
Use the rubric to evaluate your own work or ask a suitable mentor. Written assignments are not uploaded or automatically marked.