Plan the interactive to explain its essential task without a staff member, then specify what a facilitator should add. Staff can invite a hesitant visitor, help a group interpret a result, or recover from a stalled device. They can also unintentionally become the interface: if every participant needs a spoken explanation before touching the screen, the installation has an unresolved communication problem.
Museum research gives a useful, bounded way to think about the role. Pattison and colleagues' 2017 design-based study of experienced educators at interactive math exhibits describes responsive facilitation as a cycle of observing, supporting, and reflecting. It is a model for educators in family learning, not a proof that a brand booth needs one staff member per station. TERC research summary.
What does a person add to the screen?
A facilitator can notice a visitor's uncertainty and give a small prompt. A static instruction cannot adjust its wording after seeing the reaction. But a person also changes the social situation. In a 2021 study at a children's museum, 95 children aged 4–7 and caregivers were assigned to receive similar prompts from a facilitator or labels on circuit blocks. The study found little difference in who set goals between conditions; caregivers performed fewer circuit-connecting actions in the facilitated condition, while children completed more predefined circuits there. The paper notes the setting, predominantly white and highly educated sample, and that the facilitator prompted throughout rather than fading away. It does not tell us that one mode is always better. Letourneau and colleagues, Frontiers in Psychology, 2021.
Another exploratory study recorded ten families at an interactive biodiversity exhibition in Brazil. Human mediation did not change the amount or types of visitor–exhibit interaction or time at the exhibit in that sample, but was associated with more group conversation, including science-related talk. It is a small family and museum study, not an event sales result. Massarani and colleagues, 2022, via Oregon State University.
Together, these studies suggest a commissioning question: which visitor behavior is the staff member there to support? If the desired result is a short independent try, staff should not have to narrate every step. If the goal is a conversation about choices, a skilled prompt may be part of the experience. Those are Archv design inferences from the evidence, not measured outcomes for a client.
Give staff three bounded jobs
Write a one-page operating script with three modes. Invite: a brief, accurate sentence says what the visitor can do; this should not be the only way to discover the interaction. Support: an open prompt helps when someone pauses, while the visitor keeps control. Recover: staff know how to restart a stuck session, direct someone to an equivalent route, and call technical help. For each mode, say what staff must not improvise, such as unapproved product claims or a privacy promise.
Then rehearse the same core task twice with first-time testers: once with no intervention, once with the planned staff behavior. Observe starts, completions, misunderstandings, and conversation. Do not interpret a longer dwell time by itself as success; it can also mean the task was hard. Ask the visitor what they understood and ask staff which prompts they used. The interaction-discoverability guide addresses the screen's invitation; the group design guide addresses people taking turns around it.
If the unaided route fails, fix the interface. If staff meaningfully add context, staff the event and train the prompts. Record both decisions in the brief so an unplanned staffing cut does not change what the visitor is able to do.