How we run a test session with real children

What do we mean by a real test session?

Our test session is not a demo we perform for a child while recording a smile, and it is not a survey collecting ready-made opinions. It is watching a real child deal with a prototype, often unfinished, while we notice what they do before what they say. The difference between the two matters: a demo measures a child's reaction to a finished product, and a test measures where a design breaks before it reaches its final form. The kid-test stage is a declared part of our process, coming after the gray prototype and before launch, and nothing here counts as finished before it.

This article explains how that session actually runs: consent beforehand, observation during it, the questions we avoid, and how a child's word becomes an actual design decision.

Before the session: two consents, not one

The standard used in childhood research separates two distinct consents, neither substituting for the other: the guardian's written consent for their child to take part, and the child's own consent, in their own simple words, to take part specifically. Ethical Research Involving Children (ERIC), a collaboration between the Centre for Children and Young People at Southern Cross University and UNICEF's Office of Research in Innocenti, defines informed consent as the process of telling a child and their guardian the purpose of the work and their role in it before the decision to take part, one that stays open to withdrawal at any moment rather than granted once and forgotten. We hold to that distinction: a parent's consent is the first condition, a child's consent is a second, independent one, and the right to withdraw stays with the child alone, even with a parent fully willing for the session to continue.

Preparation does not stop at consent. Choosing a setting familiar to the child reduces tension that distorts observation, keeping the number of adults present to the minimum needed prevents a child from feeling like they face a panel, and setting a length that suits their age protects the session from a fatigue that produces notes reflecting nothing real about the design.

During the session: we observe before we ask

The thing most worth watching in a good session is not what a child says, but what they do before being asked their opinion. Where do their eyes go first? Do they try something themselves, or wait to be shown? Where does their hand hesitate before it touches? These signals come before any sentence a child says, and are often more honest than it, because an action is not trying to please anyone while speech sometimes does, without the child even meaning to.

With younger children, behavioral observation carries more weight than any direct question, because their ability to describe an inner experience in words is still forming. With older children and teenagers, it works to ask them to think aloud while trying something: to say what crosses their mind moment by moment, not a final verdict once the attempt is over. The difference matters: thinking aloud captures the moment itself, and a final verdict captures the child's memory of it, which is less precise.

What do we not ask?

One question ruins an entire session more than any other mistake: a question that suggests the wanted answer inside its own wording. "Didn't you like this?" pushes a child toward the answer they think will please the person asking, while an open question like "what happened here?" leaves them room to say what they actually saw. We also avoid adult design vocabulary: a child has no word for "usability," but they have an action that shows hesitation or enthusiasm, and that action is what we record, rather than extracting a technical term from a child nobody asked to learn.

More important than how a question is phrased is what we leave after asking it: enough silence. The natural adult instinct is to fill every silence with another question, while a child often needs a moment to find the right word. Interrupting that silence steals the real answer and replaces it with the first thing that comes to mind to end an awkward moment.

From observation to a design decision

An observation that never becomes an actual change in the file is a wasted observation, however carefully it was written down. In the illustrative example we published earlier about how an idea is born, a seven-year-old's sentence, "the star is small," was not filed into a report. It became a change in the size of an element on screen. That is the real test of whether any session was taken seriously: did a child's word change something concrete, or did it stay a nice sentence in a notes file nobody reopens?

Researchers in this field distinguish between different degrees of a child's participation, from being merely the tester of a finished design to a full partner in the design itself, with a rung between them where a child is consulted seriously without designing anything themselves. Our test sessions sit somewhere along that range rather than at one fixed point: how much a child participates varies with their age and the stage of the work, but the constant law is that their opinion, whatever degree of participation it came from, leaves a mark on the version that ships rather than living only in a report kept to itself.

What if the child says nothing?

Silence is not an absence of information. A child who stops trying something without complaint is saying something clear about that part of the design, even without a word. A child who repeats one attempt with visible enjoyment is saying something just as clearly. We treat stopping play, repeating a particular attempt, and ignoring an element entirely as complete observations, not as a gap waiting for a child to fill it with a spoken sentence.

Where should you start?

If you are planning a test session with children as part of your project, start with the two consents before anything else, and phrase your first question so it does not suggest its own answer. You can read how an idea is born to see where this session sits within the full path, or tell us about your project through the Start a project form.

References

• Ethical Research Involving Children (ERIC), a collaboration between the Centre for Children and Young People at Southern Cross University and UNICEF's Office of Research, Innocenti: definition of informed consent, retrieved 6 September 2026: https://childethics.com/glossary/informed-consent/
• Hart, R. (1992). Children's Participation: From Tokenism to Citizenship. UNICEF Innocenti. An academic reference with no link.
• Druin, A. (2002). "The Role of Children in the Design of New Technology." Behaviour and Information Technology. An academic reference with no link.

Quick questions

Does the child need to consent themselves, or is a guardian's consent enough?

We need both together. A guardian's written consent is the first condition, and the child's own consent, in their own simple words, is a second condition that does not fall away once the first is met. The right to withdraw at any moment stays with the child alone, even if their parent is fully willing for the session to continue.

What makes a child's observation honest rather than polite?

Asking open questions that do not suggest the wanted answer, and leaving enough silence after each one instead of filling it with another question. A child who feels an adult is waiting for their real opinion gives it; one who senses a correct answer is expected looks for that instead of their own view.

Do we tell a child when the prototype fails its task?

We do not tell them they got it wrong. We tell them their words helped us change the thing that did not work. The failure here belongs to the design, not the child, and we take care to make that difference completely clear.