Understanding Museum Audiences Through Cluster Analysis

We have seen a surge in requests for audience research projects over the last couple of years. All types of museums, from science to history to botanic gardens, are asking for studies that can help them understand more about their visitors. And with this flurry of audience research projects, we have been doing a lot of thinking about museums and their audiences. This past spring, Amanda Krantz wrote about the value of audience research to museums for building relationships and empathy with visitors. She also cautioned about pitfalls of audience research—studies that focus solely on demographics, do not truly seek to truly understand audiences, and lack a clear vision for how the information gathered will inform a museum’s  practices. And more recently, Stephanie Downey talked about the utility of audience composites for understanding your audiences’ characteristics, behaviors, goals, and pain points and creating a pathway for audience strategy.

Audience research is one of my favorite types of projects. It is full of potential for helping museums think beyond demographics and toward understanding how visitors’ motivations, values, and experiences shape their relationship with the museum and their visit. It also uses one of my favorite methods: surveys. Surveys are useful and economical for collecting large amounts of information from people relatively quickly. They also yield quantitative data which we can use to look for relationships among variables—for example, how are a museum’s local visitors’ motivations different from tourists? Or, how do visitors in social groups value certain types of museum experiences compared to solo visitors? With a large enough sample, we can start to draw generalizable conclusions about a museum’s visitors.

One tool that we can apply to survey data for audience research is K-means cluster analysis (which I’ll just refer to as “cluster analysis” for the remainder of this post). Cluster analysis is a statistical procedure that uses rating scale data to place visitors into similar groupings, or “clusters,” based on shared psychographic characteristics (such as thoughts, attitudes, values, or preferences). Rating scale questions (like the one pictured below) are those that ask visitors to choose a numerical rating for a series of statements. The poles of a rating scale can vary depending on the goal of the question: for example a scale from strongly disagree (1) to strongly agree (7) might measure attitudes, while a scale from not at all important to me (1) to very important to me (7) might measure values.

Designing the right rating scale statements is very important because they are the foundation of a cluster analysis—they should connect to the research question you hope to answer about your visitors. For example, you might be curious about your visitors’ pre-existing relationship to art; or perhaps you want to understand your visitors in terms of their visit experience preferences. Based on how visitors rate these statements, the resulting clusters can help a museum understand and conceptualize its visitors not by focusing on their demographic characteristics, but by what values and motivations their visitors share and what makes each cluster unique. 

Here is an example of what these clusters might look like, from a past project with an arboretum:

Knowing the nuances of your visitors’ values and preferences through cluster analysis can help you conceptualize your visitors by similarities that are important (that is, groupings that speak to your research questions and can be used to inform your practice). For example, with the “Social First Visitors” segment above, the arboretum might think about the ways in which the current visit experience supports (or does not support) conversation and provides places for visitors to gather.  

It can also help you spot meaningful differences. For example, in a Curator article by Kera’s Amanda Krantz with Randi Korn and Margaret Menniner, the authors provided an example from a cluster analysis for the Dallas Museum of Art where understanding the differences between clusters provided crucial insights. Two clusters both highly rated statements about engaging with art. However, there were important differences in how they rated statements about interpreting artworks on their own versus being provided with interpretation about an artwork. As the authors explain, “These two clusters are both at the top level of engagement with art, but one group doesn’t want to be told what to think, while the other group is a sponge for information.” This distinction provided the Dallas Museum of Art with a useful way to think about how the two different types of audiences might want to engage with art during a visit.

Cluster analysis is a useful tool that can help you better understand your visitors and make decisions that will help your organization build relationships, create audience-centered exhibitions and programs, and focus resources. But remember, cluster analysis is just one tool in your audience research toolkit, and the best fit depends on the questions you have, resources available, and how you would like to use the information to inform your practice! I’m excited by the opportunities ahead to help set a foundation for meaningful experiences and relationships between museums and their audiences as more museums pursue audience research.