Research Study

Origin & evolution

The discipline grew out of human factors and applied psychology after the Second World War, where studying how people actually used equipment became a formal practice. It reached software through usability engineering in the 1980s and 1990s. Jakob Nielsen's work, summarised later as the case that five users find most usability problems, reframed the study as a small, fast, repeatable thing rather than a year-long academic project, which is what made it survivable inside a product team's schedule.

The methods themselves diverged into distinct shapes. An interview study asks people to recount and explain; a usability study watches people attempt tasks☑️TaskProduct SpecificationA unit of work within a story or epicView reference →; a survey trades depth for sample size; a diary study tracks behaviour over time; an analytics study reads what people already did. Each answers a different question, and choosing the wrong method is the most common way a study fails before it begins.

Continuous discovery, articulated by Teresa Torres in *Continuous Discovery Habits* (2021), changed the cadence. The set-piece study commissioned twice a year gave way to a steady weekly habit of touching customers, which made the lightweight, well-scoped study the working unit of research rather than the occasional event.

How it works in practice

A team wants to know why second-week retention drops. They scope a study before recruiting anyone: method is interview, target is twelve participants who churned between day seven and day fourteen, the window is three weeks. The research question❓Research QuestionUser ResearchA question guiding a research studyView reference → is written first, "what did returning users expect to find and not find?", because a study without a question collects data nobody can use. Each session produces tagged observations👁️ObservationUser ResearchA specific behaviour or statement observedView reference → and verbatim quotes, which cluster into affinity groups, which synthesise into three insights🔦InsightUser ResearchA synthesised finding from researchView reference →. Six months later a teammate asking "how do we know this?" can open the study, see it was twelve churned users in March, and weigh the finding accordingly. The sample size is part of the record, not lost in a slide.

Research study vs. its neighbours

• Experiment🧫ExperimentValidationA test designed to validate a hypothesisView reference →: a study is open-ended and seeks to understand; an experiment tests a specific hypothesis🧪HypothesisValidationA testable belief about a solutionView reference → against a success criterion. A study asks "why do users churn?"; an experiment asks "does a guided import raise week-two retention by ten per cent?" They sit in adjacent domains of the same discovery region.
• Research insight: a study is the investigation; an insight is one interpreted output of it. A single study can produce several insights, which is why the graph links them one-to-many through research_study_produces_insightResearch StudyproducesInsighthierarchy rather than collapsing the two.
• Observation: an observation is a single thing seen during a study; the study is the framed effort that captured it. Observations float meaninglessly without the study that gives them a method and a date, which the graph prevents through research_study_captures_observationResearch StudycapturesObservationhierarchy.

In the graph

In the Unified Product Graph, research_study🧑‍🔬Research StudyUser ResearchA planned research activity anchors the user-research creation sequence inside the Discovery, Research & Validation region. It is the first node in that sequence, the thing everything else hangs from. Its properties, method, participant_count, start_date, and end_date, encode exactly the metadata that lets a reader judge rigour. Outbound edges build the full research chain: research_study_investigates_research_questionResearch StudyinvestigatesResearch Questionhierarchy, research_study_enrolls_participantResearch StudyenrollsParticipanthierarchy, research_study_captures_observationResearch StudycapturesObservationhierarchy, research_study_clusters_into_affinity_clusterResearch Studyclusters intoAffinity Clusterhierarchy, and research_study_produces_insightResearch StudyproducesInsighthierarchy. A study with questions but no insights is a stalled investigation; insights with no parent study are findings nobody can audit. The structure makes both states visible.