Prototype Testing With Real Visitors

The idea

A prototype is a rough, incomplete version of an exhibit, built to test an idea. Prototype testing means putting that version in front of people who resemble the intended audience, watching what happens, and changing the design based on what you learn. It happens during design, after the early audience research and before the final build.

The reasoning is about cost. A problem found on paper or in a rough mockup can often be fixed in an hour. The same problem found after programming, fabrication and installation may need weeks of work or may be left unresolved because the budget is spent. Testing early is also less personal: no one has invested months in a rough cardboard version, so changing it is easy.

What counts as a prototype

Prototypes range widely in how finished they look. Fidelity should match the question being asked.

• Paper and sketches. Hand-drawn screens that a person taps while a facilitator swaps the pages. This is enough to test whether the flow makes sense and whether labels are understood. See paper prototypes for the technique in more detail.
• Clickable mockups. Simple linked screens with no real function behind them, to test layout, wording and touch targets.
• Cardboard and foam models. Mockups of physical exhibits at full size, to test height, reach, where people stand and how a group fits.
• Partial working versions. A single interactive running on a laptop, with the rest of the experience described verbally.
• Staged “wizard of Oz” tests. A person behind the scenes simulates what software would do, such as responding to speech. This allows testing a concept before building the technology.

A common mistake is making a prototype look too polished. People are more reluctant to criticize something that appears finished, and the team becomes reluctant to change it. Rough versions invite honest feedback.

What to test

Decide the question first. Examples include:

• Do visitors understand what the exhibit is for without instructions?
• Can they find the main control?
• Do they understand the wording of the labels?
• Does the activity take about as long as expected?
• Do groups share it, or does one person take over?
• Is the content understandable for the least-informed members of the target audience?

Tests that try to answer everything at once usually answer little. One or two questions per round is typical. If your goals are written clearly, the questions can come directly from them.

Who to test with

Real visitors from the intended audiences, not colleagues, are what gives the test value. Staff know too much, and friends tend to be kind. Possible sources include people already in the gallery who agree to try something, community groups, school groups arranged in advance, and volunteers.

Aim to include a range: first-time visitors and returning visitors, children and older adults where relevant, and people with different access needs. Testing with someone who has low vision or someone who finds complex screens tiring may reveal problems that nobody else would, and many of them affect other visitors too. An accessibility checklist is a useful reference, but it does not replace seeing real people try the exhibit.

You do not need large numbers for early rounds. Even a handful of people trying the same rough version tends to reveal the major problems, since the same points of confusion often repeat. Because the numbers are small, treat the results as clues to problems, not as measures of how many visitors will struggle.

Running a session

A simple structure works well:

1. Welcome and consent. Explain that you are testing the design, not the visitor. Ask permission to take notes, and to record if you plan to, and explain how the information will be used.
2. Give a task, not instructions. For example, “You’re visiting and see this screen. Show me what you’d do.” Avoid explaining how it works. If you have to explain it, the design may need the explanation.
3. Watch quietly. Note where they hesitate, what they touch, what they say. Resist helping straight away. Wait to see whether they solve it unaided.
4. Encourage thinking aloud. “Tell me what you’re thinking” can reveal expectations. Some people find this natural, others do not.
5. Ask afterward. Open questions such as “What did you expect to happen?” are more revealing than yes/no.
6. Thank them.

Two staff members is ideal, one to run the session and one to take notes. If you have only one, record brief notes right after each person.

Do not defend the design when a visitor struggles. The aim is to see the difficulty, not to prove the visitor wrong.

Making sense of what you see

Look for patterns across sessions. If three of five people tapped the wrong area, that is a design issue. If only one person had a problem, note it, but do not rush to change anything. Group findings by seriousness:

• Blocks completion. Visitors could not do the main thing. Fix first.
• Causes confusion or delay. They got there, but with difficulty.
• Minor annoyance. Worth noting, lower priority.

Pay attention to what people did rather than what they said they liked. Visitors are often polite in conversation, and a smiling “that was great” can accompany a frustrating experience.

Iterating

Testing is meant to be repeated. A typical cycle is: test, list the problems, change the most serious ones, and test again. Each round should be quick. It is more valuable to run three short rounds than one long one, because fixes themselves can introduce new problems.

Keep a short log of what was tested, with whom, what happened, and what was changed. This is useful to the whole team, to the person writing the final documentation, and to anyone who later asks why a feature works as it does. It also feeds the project schedule: build time for testing and revision into the plan from the beginning, since squeezing it out at the end is a common reason it is skipped.

Limits

Prototype tests have limits worth stating plainly:

• A rough version in a quiet corner is not the same as a finished exhibit in a busy gallery. Noise, crowds and glare may create issues the test did not reveal.
• Visitors may behave differently when they know they are part of a test.
• Small samples cannot say how common a problem is overall.
• Prototypes rarely show how an exhibit holds up over weeks of use, so later checks remain necessary.

None of this makes testing less worthwhile. It means early testing is one stage among several, and it works best when followed by checks of the finished exhibit once it is open.