Why watch instead of ask
People are often poor reporters of their own behavior. They forget what they touched, they say they read labels they skipped, and they are polite about things that frustrated them. Watching what visitors actually do fills in the gaps. It shows the path they took through a room, where they stopped, what they ignored, and where they looked puzzled.
Observation is also one of the cheapest methods. A notebook and a clear plan can produce useful findings in an afternoon. It does not depend on people being willing to stop and answer questions, which means it also captures visitors who would never complete a survey.
What you can see
Observation is best at recording visible behavior. Typical things to note include:
- Movement. Which way people turn on entering, which stations draw them first, where traffic bunches up, and whether visitors follow the intended route. This links to wayfinding in the gallery.
- Stopping. Whether people stop at all, and for roughly how long. Timing them in more detail is a topic of its own: dwell time.
- Touching and operating. What visitors press, in what order, whether they find the controls, and whether they use them as the designers intended.
- Reading and looking. Whether they glance at a label, lean in, or read aloud to someone.
- Social behavior. Whether people go in groups, who operates the exhibit, who watches, whether adults help children, and whether strangers share a screen.
- Signs of trouble. Repeated tapping, backing away, searching for instructions, asking a companion or staff member for help.
- Barriers. Crowding around a screen, someone unable to reach controls, glare on a display, or noise from a neighboring exhibit.
What observation cannot show is what is happening in people’s minds. You can see that a visitor read a label for twenty seconds; you cannot be sure they understood it. Pair watching with brief conversations when you need meaning.
Deciding what to look for
Observation without a purpose produces pages of notes and few conclusions. Start with one or two questions, such as “Do visitors discover the second screen?” or “Where do people hesitate at the entrance?” Your goals, if they have been written in a checkable form, usually suggest the behavior to look for.
Then list the behaviors you will record, and how. A short, simple form works better than free-form notes alone. For each visitor or group, you might record:
- Time of day and rough crowd level.
- Whether they approached the exhibit, stopped, or walked past.
- The first thing they did.
- Any problem or hesitation, described briefly.
- Whether they finished the main activity.
- Approximate group makeup, for example adult alone, adults together or adults with children, without recording personal details.
Pilot the form for an hour and revise it. Forms nearly always need changes once you see real behavior.
Methods
Unobtrusive observation. The observer stands at a distance, perhaps appearing to read a label or use a phone, and records behavior without speaking to anyone. This reduces the chance that people change what they do because they know they are watched. Be open about the practice at an organizational level: many institutions post a notice that observation research takes place, and staff should know and be able to explain it.
Tracking. The observer follows a visitor from entry, noting where they stop and for how long, and sketching the route on a floor plan. Tracking gives a rich picture of movement but is time-consuming, and the visitors followed should be chosen by a rule decided in advance, such as every fifth person to enter.
Timed sampling. Rather than follow people, the observer counts what is happening at a spot at set intervals, for example every few minutes. This is efficient for finding out how busy a station is and how it is used.
Video. Cameras can capture many hours, and recordings can be reviewed more than once. They raise additional responsibilities around notice, consent, secure storage and who can view footage; visitor privacy and visitor consent both need attention. Recording is not always appropriate, and in some spaces it is not allowed.
Participant-aided observation. In some tests, visitors are asked to try a task while the observer watches, often with an invitation to think aloud. This overlaps with prototype testing and is more intrusive, but it can quickly show where an interface confuses people.
Doing it respectfully
Visitors are guests, not subjects. Some guidelines that tend to apply:
- Follow your institution’s research and privacy policies, and get any required approval before starting.
- Do not record names, faces or other identifying details unless there is a clear need, consent and a plan for protecting them.
- Avoid observing children in ways that could cause concern. Ask about local rules on research with young children and about extra care around schools and families.
- Keep a polite distance, and stop if anyone seems uncomfortable or asks what you are doing. Be ready to explain in plain terms.
- Be careful when observing sensitive content areas. Visitors may be emotional, and watching closely can feel intrusive.
Reducing observer bias
Observers see what they expect. If you hope an exhibit is popular, you may notice engaged visitors more than puzzled ones. Some ways to limit this:
- Use the written form consistently, recording both successes and problems.
- Where possible, have two observers record the same sessions and compare. If they disagree often, the categories need clearer definitions.
- Observe at different times and on different days. A school group morning looks nothing like a quiet afternoon.
- Describe what happened in plain terms (“pressed the same button four times”) rather than judgments (“seemed bored”).
- Include the awkward results in your summary.
Watching also changes behavior in small ways, and observers cannot be invisible. This is a limit to keep in mind, not a reason to avoid the method.
From notes to findings
After a session, write up the notes the same day, while memory is fresh. Look for patterns rather than anecdotes: a problem that appears once may be personal, one that appears repeatedly is likely to be built into the design. Tally simple counts, such as how many of the observed groups found the main control without help, and report them as counts from a small sample, not as percentages of all visitors.
Sketches help. A floor plan with arrows showing common paths, or a diagram of where people clustered, communicates quickly to colleagues who were not present.
Finally, decide what to change, and keep the notes. Observation done before and after an adjustment gives a fair way to see whether the change worked. Observing a rough version early is one of the cheaper ways to catch design problems, and returning to watch the finished exhibit in use shows whether the early fixes held up once real crowds arrived.