Analytics in the Arts: An Interview with Danielle Suh

“…data exists in a specific context and there are so many other factors involved, so a large part of what we do is making sure that all of those things are accounted for. Especially because we’re talking about an art form, it’s not just a sterile “this is what the numbers say and that’s what we are only driven by” scenario.”

Picture from Danielle Suh

Danielle Suh is the Manager of Strategic Analysis and Business Development at New York City Ballet. In our interview, she discussed some of the analytics work the City Ballet is doing, the challenges involved, and her thoughts on the growing presence of analytics in the arts.

Mansi: What are some of the ways that New York City Ballet uses analytics?

Danielle: Our projects at the ballet can range across a lot of different things. In terms of the analytics, it’s everything from running a survey and looking at how people feel about various programs that we’ve run, to looking at data and figuring out who is coming to the ballet. How frequently do they come, what kinds of things are they interested in? Just thinking about how we can use that to help inform future decisions. And then we also run things like programming analysis where we essentially can look at ticket sales for specific ballets, and try and pick out all of these external factors to see how we can anticipate certain ballets selling and then use that to help inform some of the decisions that the artistic staff is making.

Mansi: What kind of data is involved in these analyses, and where does it come from? Is data quality ever an issue?

Danielle: For example, if we run a survey, obviously we have the responses that people give to us. Beyond that, we also have our own internal ticket sales that are tracked through the database. We also keep track of donations coming in so then we would know for example, this patron has contributed to the ballet x number of times over x number of years. Or this audience member has bought these kinds of subscription year after year and things like that. I think in general our data sources can be kind of varied.

All data that you work with is always going to have its inconsistencies and flaws. For example, say that someone is buying tickets through our website and they have two separate email accounts associated with them – then we can’t necessarily tell that that is the same person. I think that’s why we recognize that all data will have some of those little flaws and inconsistencies. I think we do our best to account for those as we can. The other big thing for us is that I think the data is informative and helpful, but it is always just one input into a broader, more nuanced process.

Mansi: Could you talk a little more about this broader process and how analytics fits into the programming process? As a musician, I know that there is a wealth of knowledge and experiences that people in the arts build up over their entire careers, and now all of a sudden if there’s a model telling you to do otherwise, how do you balance between traditionally what’s been done by people who have been in the arts for so long versus what a piece of software tells you?

Danielle: Definitely! So we run that programming analysis every year to help inform what is going to go on stage in future years. But City Ballet is a ballet company first and foremost, and there’s an artistic mission at the core. The data might tell us there’s one specific program that might not be selling as strongly as we wanted it to, but that’s not the only thing that we’re looking at. And so [artistic leadership] will come in saying, well, you know, we think that that is a really important ballet for audiences to see, or this ballet has this historical significance, or gives our dancers these kinds of opportunities, or whatever the other factors may be. These are going to be the primary decision-making factors. It’s not just like our one model spits out this number and we will do whatever it suggests.

I think that is kind of a critical piece – that data is just one aspect of it. I’m sure as you’ve talked about in your classes, data exists in a specific context and there are so many other factors involved, so a large part of what we do is making sure that all of those things are accounted for. Especially because we’re talking about an art form, it’s not just a sterile “this is what the numbers say and that’s what we are only driven by” scenario. Ultimately, the programming is decided by the artistic director, and they make that final call.

Mansi: One of the things that we hear about a lot is gaining support from higher leadership to support analytics activities because data science is such a new field. Has that been your experience at the New York City Ballet, especially since using analytics in arts organizations is so new?

Danielle: Yeah, I definitely hear that is a real concern. I feel very lucky that at City Ballet, people are very bought-in to these projects. The fact that I do get brought on to these projects is indicative of broader acceptance of the analytics. I’m never really asked onto projects where people don’t want input from data, and just the way that City Ballet is as an organization, generally people are very open to it.

I feel like so much of that goes to what you’re saying about building those relationships and bringing people along on that journey and that you are not just marching in being like, “I know all these things about the way you guys run that you guys don’t know.” It really is more of a collaborative process. Even in the process of doing these projects you have to figure out: what kind of data do you have access to? Do you have any concerns about the validity of this data? What do you want to use this for? You know, there are so many steps in the process where you can get people bought in and really bring them along with you. And I think all of that makes the adoption of analytics later on a little bit easier.

Mansi: You mentioned some of the different types of programming analysis and customer analysis being done at the City Ballet. Have you seen any results that have surprised you?

Danielle: I don’t know if there are specific surprises but there are little things where even if it’s not surprising, it is kind of fun to have it quantified, right? It’s like you kind of know in your head that Saturday evening performances are probably more popular than Tuesday evening performances, but it’s always cool to see that backed up in the data. We know for example that our winter season generally sells better. If I were an audience member, I don’t know that I am thinking, “oh, it’s winter time, it’s ballet time”. As an audience member, I don’t necessarily see, “oh, the winter, it’s so much more popular” or anything like that. (To be clear, we’re not talking about the Nutcracker, but our winter season is in the January, February time frame generally.) So when you do see those trends in the data, it’s always like, “oh, that’s interesting, I never would have guessed!”

We have our hypotheses maybe that, you know, when the weather isn’t great you’re less likely to have people just hanging out and sitting in the park. Instead they might choose to hang out in a theater, which might be a little bit more climate controlled. But we have limited resources, limited time. I think that it gets into a question [of] which kind of surprises or anomalies that we really want to get into versus [what] we just kind of accept so our time is better spent elsewhere.

Mansi: Do you see arts analytics like this growing as an application of the analytics field, and where do you see it progressing?

Danielle: Yeah, I do think you are seeing more arts companies and organizations realizing that this can be a powerful tool. I think to your point, because it is so new, I don’t think that we are as sophisticated as we could be. For example, at City Ballet, we’re not doing any kind of  big data analytics or machine learning or anything like that. But I think in general you’re seeing more positions like this becoming available or generally hearing people talk about practical applications for this kind of analytics, and that makes me feel like it’s going to become a bigger thing from here on out. I think it’s always very exciting because a lot of smart people are doing interesting analytics-related things in the world and there’s a hope more of them can get into the arts and help these institutions that are so valuable do well; it’s always very exciting!

Mansi: Finally, do you have any tips or advice for analytics professionals wanting to apply their skills in the arts?

Danielle: I think that the biggest thing is just being able to explain to people why your skillset would be helpful. Again, I feel very lucky at City Ballet that our executive director and everyone else in the organization was very on board and understood how we see this skill set being helpful here. With analytics in the arts being a somewhat newer thing, being able to articulate the kinds of projects you think you can help with, how we have found data to be valuable, or how an organization can be helped by your skill set [is important]. I think just having a very clear explanation for that is helpful.

When you’re coming from the analytics background, it’s so obvious to you because you spent all your time in it and everyone around you is also doing it and understands what it is. But I think that the arts can be such a different world. Ultimately the arts is not only a numbers thing, you know? So, just realizing that this is a different world and [being] able to kind of translate in your own words from one world into the other – that communication piece is very helpful.

Columnist: Mansi Shah