The starting point
Measure engagement by defining the audience action and intended outcome first. Useful evidence may include completed interactions, viewing time, waiting, ease of participation and voluntary feedback. Keep those measures distinct from whole-event attendance, economic effects or long-term changes that have not been evaluated.
Write the question you want the evidence to answer
A commission seeking an understandable interaction needs different evidence from one exploring a learning idea or a visitor journey. Start with a short question: can visitors identify how to begin, do they complete the interaction, or can they explain what they changed? Selecting the question first prevents an easy-to-collect number from becoming a substitute for the actual purpose.
Define the unit before counting
A viewer, participant, interaction and visit are not interchangeable. One person may return several times, while a group may watch a single participant. Decide what your count represents and record the method alongside the result. If you cannot identify unique people, describe the figure as observed interactions or visits rather than inventing a unique audience total.
Observe the encounter at useful moments
Consider the approach, start, response and exit. Note whether people recognise the invitation, ask for help, wait, leave before starting or repeat an action. Sample different operating periods where practical, and record when observations were made. The purpose is to understand the encounter, not to assume that the busiest moment represents the whole presentation.
Pair activity with an appropriate question
A short voluntary conversation can add context to an observed action. Ask what a participant noticed, changed or found difficult. If a learning objective was agreed, use a relevant prompt rather than a broad satisfaction score alone. Make the collection method proportionate and discuss information handling with the responsible team. A response from a small sample should not be presented as everyone’s experience.
Record the operating context
Opening hours, interruptions, changed conditions and facilitation can affect what you observe. Keep a simple session log so the team can interpret the audience evidence alongside the actual presentation. Where a baseline comparison is useful, agree a relevant period and method in advance. A comparison without context may describe a difference without explaining why it occurred.
Report the result at the level the evidence supports
Completed interactions can demonstrate activity. Feedback can describe what respondents said. Neither alone establishes spending caused by the artwork, enduring belonging or a clinical benefit. Keep intended outcomes, observations and interpretations separate in the report. The supplied case studies include suggested measures for future commissions; those suggestions are not retrospective results from the original works.
Common questions
Can we use the festival’s total attendance as our artwork’s audience?
Not unless the evidence genuinely establishes that relationship. Report event attendance separately and describe the installation-specific measure and method you have available.
Is a long dwell time always a good result?
It needs context. Time may reflect interest, waiting, confusion or another condition. Interpret it alongside the interaction design, observations and feedback rather than treating duration as a universal quality score.
When should evaluation be planned?
Before delivery, while the team can still agree the questions, responsibilities, collection method and useful baseline. Build a realistic plan into the brief rather than relying on information that may not exist after the event.
Explore the projects
Using this guidance
This article proposes an evaluation approach for a future brief. It does not report results from Georgie’s past commissions. The linked case studies identify their documented work and suggested future measures.
Read next
Prepare a conversation
Use the public briefing checklist, explore interactive public art and projection mapping, then request a concept call. Start with your site, audience and intended experience.