Contextual inquiry

012 min

The gap contextual inquiry closes

Picture a product manager at a company that makes handheld scanners for warehouse workers. They interview six pickers, the people who collect items from shelves to fill each order. All six describe the same routine: scan the shelf label, scan the item, confirm the count. The team builds a faster version of that routine, and the pickers still make mistakes. This scenario is illustrative.

A visit to the warehouse would have shown something the interviews did not. A picker keeps a paper list in a pocket and checks it before each scan, because the scanner's list scrolls too slowly. Nobody mentioned the paper list, because nobody thinks of it as part of the job. It is simply how the work gets done.

Hugh Beyer and Karen Holtzblatt designed contextual inquiry because surveys and interviews depend on what people remember and can explain. When people summarise their own work, they leave out their reasons, their motives and the mental models behind each step. A researcher who watches the work, and asks about it as it happens, collects those details first-hand.

The method works in the field, which means research done where the work happens: a home, an office, a warehouse floor. The researcher sees real tasks, in real surroundings, with real interruptions. That is the whole idea, and the rest of this article is about how to arrange it so the participant can teach without performing.

021 min

Where contextual inquiry came from

The method has a paper trail. Whiteside, Bennett and Holtzblatt first called it a phenomenological research method in 1988, which means a method that studies how people experience their own work. Wixon, Holtzblatt and Knox published the first complete account of the method itself in 1990. Holtzblatt and Beyer gave their fullest description in 1995. In 1997 Beyer and Holtzblatt published the book Contextual Design, and the book places the inquiry inside a larger process called contextual design. The inquiry is the research part of that process.

Why guides give different dates

Guides disagree about when the method began. Some say 1990, and others say 1997 or 1998. These dates describe different things. In 1990, Wixon, Holtzblatt and Knox published the first complete account of the method itself, and that account was a paper. The first edition of the book is dated 1997, and some citations of the book use 1998. One widely read guide calls Contextual Design a 1990 book, which mixes the paper and the book together. When you cite the method, name the work you mean.

031 min

How a contextual inquiry session works

Beyer and Holtzblatt chose an old relationship as the model for a visit: a master craftsman and an apprentice. In this model the participant plays the craftsman and the researcher plays the apprentice, who learns tasks by watching and asking questions. The participant is the expert on their own work. The researcher is there to learn the work, not to judge it or to teach a better way.

Four principles keep a visit on course.

• Context. The researcher goes to the place where the work normally happens, not to a lab or a meeting room.
• Partnership. The researcher and the participant explore the work together, and the participant can steer the conversation.
• Interpretation. The researcher says what they think they saw, and the participant corrects it, so both agree on what it means.
• Focus. The researcher knows what the study needs to learn, which tells them which parts of the work to watch closely.

A session usually has four parts. In the primer, the researcher introduces themselves, explains what they hope to learn and asks the participant to correct wrong guesses. In the transition, the researcher explains how watching and interrupting will work. In the contextual interview itself, the participant works while the researcher watches and asks short questions about what just happened. The wrap-up comes last: the researcher says what they understood, and the participant corrects it.

051 min

After the visit: interpretation sessions

A visit produces notes, sketches and photos, but the meaning comes afterward. In the method's interpretation session, three to eight teammates listen as the researcher retells the visit in the order it happened. The team adds its own insights and facts as notes. Retelling in order stops the team from jumping to a conclusion about one detail before they know what came before it.

The notes are then sorted. Many teams group them with an affinity diagram, which collects notes into themes on a wall or a board. Teams also draw work models, which are diagrams of how the work flows between people and tools. Notes work best when they record the sequence of steps, the materials in use, the constraints and the participant's own explanation together, so a finding keeps its setting. In the scanner example, the paper list would appear as a step in the flow that the scanner team had never drawn.

061 min

How many visits contextual inquiry needs

The method trades breadth for depth, so a team gets a detailed picture of a few people instead of a rough picture of many. A single study can involve 4 participants for one small task, or 30 or more. A session lasts from an hour or two up to several days, depending on the work being studied. Each visit also costs travel, and the team needs a few more hours to interpret the results of each one.

That cost explains when the method pays off. It suits work you do not yet understand, such as a workflow inside an unfamiliar profession. It suits a team that is about to decide what to build for that work. It fits poorly when a quick answer will do.

A remote version is possible. The participant shares their screen and works through real tasks while the researcher watches. The researcher loses the physical surroundings, such as interruptions and the layout of the workspace, but still sees the real workflow instead of a description of it.

072 min

When contextual inquiry is the wrong tool

Contextual inquiry is built for complex systems, long processes and expert users. It adds little when the task is narrow, such as redesigning a product page or testing a sign-up form. For those jobs, direct observation is enough. In direct observation the researcher watches mostly in silence and asks few questions. The method also fits poorly when the question is about attitudes, or when a decision cannot wait for several visits.

081 min

Contextual inquiry compared with other methods

Contextual inquiry is a form of ethnographic field study, which is observation of people in their own settings, reduced in scale for design work. Teams reach for it when they might otherwise choose a different method, so the table shows what each one gives you.

The fact that decides between them is whether you need to see work that nobody prompted. If an account is enough, interview. If you need to see how the work is really done, visit. For the wider family of observation methods, see design ethnography.

091 min

Using contextual inquiry on a product team

The decision this method changes is what a team builds first. Before designing a tool for work that people already do, ask whether the team knows how the work is done today without the tool. If the honest answer is no, plan visits.

1. Write the question the study must answer, so that everyone knows what to watch for.
2. Recruit people who do the work, with different levels of experience.
3. Agree on recording, bystanders and confidential material before the visit.
4. Watch first. Ask when something is unclear, and save the question about why for a natural pause.
5. Hold an interpretation session soon after the visit, while the details are fresh.
6. Check each site's workarounds against the other sites before building anything for them.

Start with a few visits on the single task you understand least. If what you see matches what people told you, the interviews were enough. If it does not, you have just found out what skipping the visits would have cost.

?3 questions

Questions people ask

§7 sources

Sources