Contextual inquiry is what you get when you cross an interview with a field observation and let the participant lead. Developed by Hugh Beyer and Karen Holtzblatt as the front end of their Contextual Design process, it is one of the most reliable ways to understand how work really happens: not the official process in the training manual, but the real one, with its workarounds, interruptions, and unwritten rules.
Where a standard interview asks people to describe their work from memory in a meeting room, a contextual inquiry goes to the work and watches it unfold, asking questions along the way. That single change addresses the oldest problem in research: asked from memory, people describe the official version of their work, not the one they actually do.
What a contextual inquiry looks like
A session typically runs from ninety minutes to a few hours, at the place where the activity actually happens: an office, a ward, a van, a kitchen. The participant does their real work, on their real tasks, while you watch, take notes, and ask about what you see as it happens. It is not a silent observation, and it is not a conversation that happens to be on site. It is a structured mix of the two, anchored to concrete work unfolding in front of you both.
That anchoring is what separates it from a plain interview. In an interview, you get the participant's summary of their work: tidied, generalised, and missing everything they no longer notice. In a contextual inquiry, the work itself keeps the account honest. The phone rings, the spreadsheet misbehaves, the handwritten list beside the keyboard turns out to carry the real process. You ask about each in the moment, while the detail is alive rather than reconstructed.
The master-apprentice model
Beyer and Holtzblatt's central metaphor is the master-apprentice relationship. You are the new apprentice; the participant is the master craftsperson. An apprentice does not hand the master a discussion guide. They watch real work, ask about what they see, and let the master explain while doing, because that is when experts are best at articulating what they know.
The stance matters as much as the structure. Approaching someone as a master of their own work, rather than a subject to be studied, changes what they show you. It grants them the authority to correct you, teaches you their vocabulary, and surfaces the pride and frustration embedded in how they actually do things. Most people, it turns out, enjoy teaching their work to a genuinely curious learner.
The four principles: context, partnership, interpretation, focus
Four principles keep a contextual inquiry rigorous rather than just a pleasant site visit.
- Context. Go to where the work happens and stay anchored to real, current instances. "Show me the last one you did" beats "what do you usually do?", because summaries hide the detail you came for.
- Partnership. You and the participant examine their work together. You are not administering a protocol at them, and you are not making small talk. You alternate between watching and discussing, and they share control of where the session goes.
- Interpretation. Observation gives you facts; design needs meaning. Share your interpretations out loud as they form: "it looks like you keep that list because you do not trust the system?" A good participant will correct a wrong reading immediately, and a confirmed one becomes solid ground to build on.
- Focus. Go in with a clear project focus that steers what you attend to, and declare it to yourself in advance. Then hold it loosely enough that a surprise can pull you somewhere better. Focus without openness misses discoveries; openness without focus produces a pile of anecdotes.
Running a contextual inquiry well
A shape that works in practice:
- Prepare a focus, not a script. A one page statement of what the project needs to learn, plus logistics: permissions, safety rules, what you may photograph.
- Open conventionally, briefly. Ten or fifteen minutes of introductions and background, then move to the work. The transition line is simple: "can you carry on with what you would normally be doing, and I will ask questions as we go?"
- Stay concrete. Whenever the participant drifts into generalities, steer back to the instance in front of you. Retrospective accounts of recent, specific events are the next best thing when live work is not available.
- Capture artefacts. Photograph forms, screens, cheat sheets, and workarounds, with permission. They anchor your analysis and your storytelling later.
- Interpret in the room, analyse across sessions. Check readings live, then consolidate afterwards; thematic analysis works well across transcripts and field notes. Three to six sessions per role is a common range, and the logic behind that number is the same saturation logic as any qualitative sample size decision.
If you want the wider family of in-context methods, from shadowing to full field studies, our guide to ethnographic research maps the territory this method sits in.
Let the work do the talking
The quiet power of contextual inquiry is that it stops asking people to be accurate historians of their own behaviour and lets the work speak for itself. Be the apprentice, stay anchored to real tasks, say your interpretations out loud, and hold your focus loosely. Do that, and you will walk out knowing things about the work that the people doing it had forgotten they knew.
People of Research is a community and newsletter for UX researchers, ethnographers, and anyone who learns best by watching real work happen. Join us and get the good stuff in your inbox.