What dwell time is
Dwell time is the amount of time a visitor spends at an exhibit or station. It is one of the simplest things to measure in a gallery: someone starts a stopwatch when a visitor stops and ends it when they walk away. Because it is simple, it is popular, and because it is popular it is often over-interpreted.
The figure can be useful. It can show whether people stop at all, whether a station holds attention for a reasonable length, and whether a change to a design shifted behavior. What it cannot do is tell you what the visitor was thinking.
How it is measured
There are a few common approaches.
Manual timing. An observer, with a stopwatch or a simple form, records when each visitor arrives and leaves. This is cheap and flexible. It also allows the observer to note who the visitor was with, what they touched and what happened next, which links naturally to visitor observation.
Video review. Recordings can be reviewed and timed later. This allows careful checking but raises questions about notice, consent and storage, discussed in the articles on visitor consent and visitor privacy. Do not record without clear signage and a policy.
Sensors and logs. Interactive screens can record when a session starts and ends. Other systems use cameras or counters at doorways. These produce a lot of data cheaply, but each has its own definition of “dwell” and its own errors, so check how the system decides when someone has arrived or left.
Whichever method you pick, write down your definition. Does timing start when the visitor stops walking or when they first touch the screen? Does it include time spent reading a nearby label? Does it end when they turn away, or when they step out of a marked area? Small differences in definition change the results.
Choosing who to time
Tracking every visitor is rarely practical. A common approach is to choose people by a simple rule, such as timing the first visitor to arrive after the previous person finishes, or every third person to stop. A rule written in advance helps avoid the unconscious habit of picking people who seem interesting or easy to follow.
Decide how to treat people who walk straight past. Some studies report the proportion who stop, and then dwell time only among those who stopped. The two figures answer different questions: the first describes pull, the second describes holding time. Reporting just an average of those who stopped can hide the fact that most people ignored the exhibit.
Also note the conditions. A crowded Saturday and a quiet Tuesday morning will give quite different results, and so will the presence of a school group. Record the date, time and approximate crowding.
What dwell time can suggest
Used carefully, it can offer clues:
- Whether the exhibit attracts and holds attention. If nearly everyone leaves within a few seconds, something may not be working, such as unclear purpose or an unresponsive screen.
- Where the time goes. If the typical stay is much shorter than the time needed to complete the core activity, people may be abandoning it.
- Differences between versions. Comparing time before and after a change, under similar conditions, can show whether the change had any effect.
- Differences between audiences. Families may stay differently than adults on their own.
- Fit with the gallery. If a station is meant to be a quick moment and people spend a long time, or the reverse, the design intent and reality may differ.
What it cannot tell you
A long stay can mean many things. The visitor may be absorbed, or confused. They may be waiting for a friend, resting, sheltering from the weather, or reading something unrelated. A short stay can reflect a well-designed, efficient exhibit that made its point quickly. An exhibit with a clear message may well hold people for less time than a puzzling one.
For this reason, more time is not automatically better. Time alone does not show understanding, enjoyment or change of attitude. If your goals involve what visitors learn or feel, dwell time is at best an indirect hint, and you need other evidence, such as interviews or brief questions. The wider topic of learning outcomes involves more than time.
Other cautions:
- Averages mislead. A few very long visits can pull up the average. The median, the middle value, is often more representative. A table or a simple list of times can show the spread more honestly.
- Social settings. A group may stay together at a station because one member wants to, while others lose interest. One person’s turn at a shared screen affects everyone’s timing.
- Observation changes behavior. People who notice they are being timed may stay longer or leave early.
- Different stations are not comparable by time alone. A video with a fixed running time will naturally have a different pattern from an open exploration of a map.
A simple worked example
Suppose, hypothetically, an observer times six visitors at a station and records stays of 20, 25, 30, 35, 40 and 150 seconds. The sum is 300 seconds, so the mean is 50 seconds. The median is the average of the middle two values, 30 and 35, which is 32.5 seconds. The single long stay raised the mean well above what five of the six people did. Reporting only the mean would suggest that visitors typically stay for close to a minute, when most stayed for under 40 seconds. Six visitors is also far too few to draw firm conclusions, so these numbers show the arithmetic only.
Reading the numbers with other evidence
Pair dwell time with something that adds meaning:
- What people did. Note whether they touched, read, talked or left. A visitor who watches a whole video, then discusses it with a companion, behaves differently from one who taps repeatedly without waiting for a response.
- What people say. A few short exit questions can reveal whether a long stay was engaged or frustrated.
- What the logs show. Where the exhibit is digital, session data can show which parts were used, though exhibit analytics need careful reading too.
If a change is made to the exhibit, measure again under comparable conditions. A single result is a snapshot; a pattern across days and visitor types is more convincing.
Setting expectations
Before collecting any data, decide what you would consider a good or poor result and why. Without that, any number can be interpreted to support whatever the team hoped. A reasonable aim is something like “most people who stop at the station should have time to complete the main activity”, which is grounded in how long the activity actually takes, tested during rough-version trials. This is a better benchmark than comparing with a figure from a different museum, where the exhibit, the audience and the layout all differ.
Dwell time is a convenient starting point and a poor stopping point. Treat it as one piece of evidence, record how it was collected, and be honest about what it leaves unanswered.