Ethnography: Uncover Needs Users Can't Articulate

WHY IT EXISTS Surveys and interviews are limited by what users can remember and articulate. People often say one thing but do another, creating a gap between stated beliefs and actual behavior. Ethnography was developed to close this gap by studying people in their natural context, revealing needs they may not even know they have.

THE MENTAL MODEL Think of yourself as an anthropologist studying a 'tribe' of users in their native habitat—an office, a factory, or a home. Your goal isn't to ask direct questions but to immerse yourself in their world. You observe their workflows, challenges, and social dynamics to understand their culture and problems from their point of view.

HOW IT WORKS A researcher embeds with participants for an extended period. The core activity is participant observation, where the researcher watches and sometimes joins in the participants' daily routines. This is supplemented with contextual interviews and analysis of artifacts (like tools, documents, or physical spaces). The researcher systematically records observations to identify patterns, interpret behaviors, and build a holistic understanding of the group's culture and unmet needs.

WHEN TO USE IT Use ethnography for foundational research at the very beginning of a project, when you're exploring a new problem space or a poorly understood user segment. It excels at discovering the 'unknown unknowns'—deeply ingrained needs that users cannot easily express. It's also valuable for diagnosing why an existing product is failing or being used in unexpected ways.

WHEN NOT TO USE IT Avoid ethnography for quick, quantitative answers. It is not the right tool for validating a specific feature, A/B testing designs, or measuring market size. Because it is time-consuming, expensive, and relies on small sample sizes, it's unsuitable for generating statistically significant data.

ONE CANONICAL EXAMPLE To improve a spreadsheet application, a survey might ask 'What new features do you want?'. An ethnographic study, however, would involve a researcher spending a week in an accounting office. They might observe accountants manually cross-referencing numbers between two printed spreadsheets, a tedious and error-prone task. The users would never ask for a 'data-linking feature' because they've always done it manually. The observation uncovers the real, unarticulated need.