Research Question

Origin & evolution

Framing the question first is borrowed wholesale from empirical science, where the research question precedes the method and the method is chosen to answer it. Product and design research adopted the same order. Erika Hall made the distinction its sharpest in *Just Enough Research* (2013), arguing that the largest source of confusion in design research is mistaking the research question for an interview question. As she puts it in Research Questions Are Not Interview Questions, you may want to know how likely people are to use your product, but you cannot find that out by asking them how likely they are to use your product. The research question states what you want to learn; the interview question is an indirect instrument you design to get at it without contaminating the answer.

The second lesson is about leading. A question that smuggles in its preferred answer ("How much did you love the new onboarding?") guarantees a useless result, because participants oblige. Good research questions stay neutral and open, and the study design, observation👁️ObservationUser ResearchA specific behaviour or statement observedView reference →, story-based interviewing, log analysis, is then chosen to answer them without prompting. Teresa Torres's story-based interviewing in *Continuous Discovery Habits* (2021) is one such design: collect specific stories of past behaviour, because what people did is better evidence than what they predict they will do.

The order between question and hypothesis🧪HypothesisValidationA testable belief about a solutionView reference → is worth stating plainly. The question usually comes first; the hypothesis is a candidate answer you hold once you have enough prior evidence to bet on one. A team that opens with a hypothesis and no question often skips the part where it asks whether the problem is even real.

How it works in practice

A team notices trial users drop off after creating a first project. The tempting move is a survey: "Was the project setup confusing?" That is a leading interview question dressed as research. The research question underneath it is broader and neutral: "What stops trial users from returning after their first session?"

That question shapes a study. The team runs eight story-based interviews asking people to walk through the last time they set something up and abandoned it, and pairs the transcripts with drop-off logs. The finding is not the assumed confusion at all; users left because the first project felt disposable, with nothing pulling them back the next day. Because the question was about why people did not return, not about whether setup was confusing, the study had room to surface the real cause and to generate concrete follow-up work.

Research question vs. its neighbours

• Hypothesis is a proposed answer with a direction you can test ("users churn because there is no return trigger"). The research question is the open enquiry that comes before you are entitled to that bet. A study investigates a question; an experiment🧫ExperimentValidationA test designed to validate a hypothesisView reference → tests a hypothesis.
• Interview question is the wording you put to a participant. It is an instrument serving the research question, deliberately indirect, never a verbatim restatement. One research question typically spawns several interview questions across a guide.
• Insight🔦InsightUser ResearchA synthesised finding from researchView reference → is what a study produces once the question is answered, a durable, evidenced understanding. The question is the brief; the insight is the deliverable. A research question with no insight against it is a study that never closed.

In the graph

In the Unified Product Graph, research_question❓Research QuestionUser ResearchA question guiding a research study anchors a study in the user-research region. research_study_investigates_research_questionResearch StudyinvestigatesResearch Questionhierarchy records what a given study set out to learn, so a study with no question is a method without a purpose. research_question_addressed_by_insightResearch Questionaddressed byInsightcross-domain closes the loop, linking the enquiry to the evidenced finding that answers it. research_question_generates_taskResearch QuestiongeneratesTaskcausal captures the downstream consequence, that answering a question often spawns concrete work, and user_advisory_board_surfaces_research_questionUser Advisory BoardsurfacesResearch Questioncross-domain records that questions also come from standing customer relationships🫶Customer RelationshipBusiness ModelA type of customer relationshipView reference →, not only from the team's own curiosity. Modelling the question as its own node keeps research honest: an open question shows what the team still does not know, and an unanswered one stays a visible gap on the board until someone closes it.