Skip to content

Leading the data agenda with Baku Hosoe from The Met

Setting up a data function in your visitor attraction? Join Baku Hosoe, Head of Data and Analytics at The Metropolitan Museum of Art in New York, to hear about the origins and evolution of The Met’s data program and Baku’s takeaways for structuring, hiring and enabling a data team.


Angie: Over in the past few years, the world of data and visitor attractions has come a very long way. So joining me today for our next episode of the Data Diaries is one of its leaders of data analytics doing that work on the ground in one of the world’s biggest cultural institutions I have with me, Baku Hosoe, the head of data analytics at The Metropolitan Museum of Art in New York. Welcome Baku!
Baku: Thank you for having me.
Angie: Baku, you came into the museum field from a consulting background. How does analytics and visitor attractions compare to what you’ve seen in industries like finance, telco and pharma?
Baku: Yes. It was a very different from what I was accustomed to. In many ways it is much more difficult to think about analytics in visitor attractions perhaps because when we think about people’s decision to visit a museum, it is not a transactional thing, it is something that’s deeply emotional and many of our visitors come with high expectations, leave with great satisfactions and often repeat. And so it is even more important than ever, and in a way, very difficult to understand the rich and nuanced experience of each of our visitors and derive insights on how we can improve.
Angie: And it’s always sat pretty close to the executive, the data function at the Met – and from memory at your board level as well. What was some of the origins and the drivers that were behind the data initiative?
Baku: It has certainly been a multi-year effort. There has always been strong interest from all around both internal leaders, as well as board members and even some of the donors have also pinged the institution about what we’re doing with data and how we’re improving the use of data. They all felt the need to invest in this area, which actually helped create my position as well.

It can be seen as a bit of a reflection to go along with the changes in board structure as well. One of the key changes they have made in recent years was to form a revenue committee, which is really aiming at improving the visitor experience all around. And that goes hand in hand with the creation of the data team to have visibility across all the data silos and business unit silos and provide the best in class experience across all the aspects of the interaction with the Met to our visitors, because ultimately whether they show up at the stores or visit the Museum or go to the concert or go on to become members and participate in different live events, they’re all the same people. And it is really important to have a holistic view about each and every one of them. 
Angie: And it’s such a wonderful thing to see, data championed at such a high level. I think it’s one of the key factors for success that we’ve seen in visitor attractions is that it really does need to come from the C level, if not even higher, and so fabulous that the whole board, is on board so to speak, at that level at the Met.
How has your role  evolved over the course of the last few years?
Baku: For one thing to state the obvious, COVID has thrown a wrench into our plan for sure. I have a stepped away from data to work on COVID related project management around the closure and reopening. But putting that aside I would say that my overarching responsibility has remained the same, which is to be the change agent and in a way, an evangelist for the use of data to derive the insights and to inform decisions and actions throughout the museum. But tactically speaking for the first year, I’d say much of my activities or around building things, creating reports, doing things for people. And compared to that, lately things shifted more to influencing or enabling people throughout the organization. That is definitely a one noticeable change that has happened.
And at the same time I think some of the key priorities or focus area have widened. It hasn’t shifted, but just widened, in that for the beginning, it was really important to show visible and measurable ROI through the pure focus on the revenue and audience. But now there are much more efforts around how to improve the operations internally through the use of data, how best to utilize the rich collections data that we have and how to establish the governance structure and a community mentality fostering  the data culture throughout the Met.

These are all the areas that have become important almost if not more than the revenue from.
Angie: That last piece is so interesting to me, the way that the focus has widened over the years, because in many cultural and institutions that we see, the focus has actually gone the other way recently with COVID, that they had started out a few years ago with a really broad scope with data. And we tend to encourage people to get narrow in the first instance, when they’re starting to work with data, but in sort of the post COVID months that we’re in at the moment, and that has become really hyper-focused on revenue. Which perhaps, for a lot of cultural institutions, isn’t something that they naturally would’ve started out with. So it was really interesting to see that your journey’s almost taken you in another direction at the Met over the past few years.
Baku: Yeah, it is certainly dependent on the institution. And I think what is happening at the Met in particular is that with the COVID and with the macro trends around the labor force, we are all asked to do more with less.
And when we think about that, it is very important to be efficient, effective, and really be able to prioritize and focus on what matters. There are technologies out there that help us do it, but there wasn’t as much impetus on required being required to do so. And COVID has changed the formula. And I think people are much more excited about having a better way to improve the way we work and the way we manage our work through data that is a driving force behind the broadening of the scope, so to speak.
Angie: So there’s an efficiency layer to that as well as there.
Baku: Absolutely.
Angie: And you make a really good point around data governance, right? That is also a piece that we see as being a really, really important part of the sort of best practice equation and cultural institutions. Because a lot of the work of data is finding out where it came from and making sure that the critic business rules are applied and making sure everybody knows what those things are. And then managing for things like privacy and security and the validity of data as well, and its integrity. Is that what that work comprises for you at the time?
Baku: Absolutely. It really is about having a more explicit structure and also a centralized focus on those areas. All the industrial changes, regulatory changes that are happening around us, impact all of us and influence how we think about collecting, storing, using data. And yet in the past, many of the data users within the Museum were operating more in silos. And there were at subscale when it comes to thinking about all the global changes that are happening. So it was all the more important to create a governance structure and committee that actually look at these things and ensure there is the right level of access, for the right reasons. And keeping transparency and above all, staying in compliance with regulatory changes… an environment, that’s constantly in change.
Angie: And one of the things I think is really interesting. You’ve you’ve built your data department up to include a few people and cross-functional roles as well from across the Museum.
I know a lot of other attractions are really busy hiring data roles at the moment, and it might include their first data roles, or somebody like yourself. But, in your view, what sorts of roles are most useful to hire and what order should they hire?
Baku: I personally think there’s no one right answer that fits for everyone. As we think about data as a function, it is still in the early stage of maturity curve, unlike other established business functions like HR, finance or IT. And I think for many of the institutions, the resource limitations, and really the degree of ambition surrounding the data topic, dictate how best to structure.
Angie: So Baku, coming back to this org structure of how you approach data, what are your thoughts on outsourcing versus insourcing for that function? One of the sort of elements that go into that decision?

Baku: Again, I think there’s really no one right answer that fits all in this particular one. One of the things that I needed to do early on as I started my role was to really evaluate our internal capabilities that existed at the time and how data was structured. And what we found for ourselves was that our data resources existed. But they were scattered throughout the organization and the types of tools they used, the level of capabilities when it comes to data topics, were all at different levels. And more importantly, there wasn’t any conversation across that data resources, because they’re all reporting into different organizations. So when thinking about the design overall, the two main questions we asked was: one, whether to centralize or decentralize data capabilities and two, how much to insource versus outsource. And I think the question is applicable to any institutions who are thinking about the design, but the right answer would be very different depending on both the reality of the organization, as well as the resource limitations and ambitions for the data function in the institution. On the questions or centralization versus decentralization, it really comes down to how different are the types of insights that are needed to be effective for particular areas. And what we found was that there is a very significant difference between say, a retail department versus a fundraising department, versus membership department. So the way we’re approaching it is that there is still a component of centralization, through my team sitting at the data analytics office, but we also have very specialized data resources in some of the key departments. And we all work together in the area where the consistency makes sense, but otherwise have developed very focused capabilities on how best to deliver insights for that particular department. And similarly on the questions of insourcing and outsourcing, outsourcing works effectively when there is not enough in sort of in house capabilities to build it from scratch. Whereas insourcing works flexibly when you already have resource who can handle that. So it does really come down to the scale of your organization and the resource availability as well, too, to decide whether to invest in particular area or not. And for us, we are definitely doing the mixture of it. So in some of the areas, we have more than one data analysts who are really going deep into to utilizing AI, machine learning, to deriving insights themselves in an in sourced fashion. In some other areas, we utilize outsource vendor for both the creation and analysis of the data. So there really isn’t one solution fits all, but really dependent on the reality of your organism.
Angie: It’s great to hear that, that such a conscious decision for you, because I think this is such an important question that is often really skipped over in favor of just talking about the solutions, rather than being a strategic one. And one of my favorite pieces of advice that I was given a long time ago on this question was to insource or to build the things that are unique to your organization. And then to outsource, or to buy the things that are common in your industry, so that you sort of concentrate your investment into the areas that make your organization special. And it sounds very much similar to some of the things that you’ve been doing at the Met. Because the total cost of ownership of some of these data solutions can be… when everything is built in house… can become incredibly, incredibly heavy to bear for one organization, when you think about maintaining something, and as, as you’ve mentioned before, going through a roadmap of, of developing something over over a period of time as well.
Baku: I cannot agree more with your statement there. One of the things that we have to realize is that as, as cultural institutions the Met is obviously one of the bigger players in the space, but even we are subscale for many of the things. And when we think about the industry at large, there’s a tremendous value in having an outsource vendor who can scale more effectively for the type of topics that impact and influence everybody as you noted.
Angie: Whereas something like collections data is going to be so unique for each organization. You know, it’s something that differs so broadly between say an art museum versus a history museum versus a science museum. And, you know, at that point there is sort of no productized approach to it is there.

And what about the sort of attributes and skills we should look for in data hires. What do you see as being the best people to get for the job?
Baku: I think, generally speaking, there are two elements to think about. One is about the business skills. And the second one is about technical skills. And really it is a question of, what’s the status with data champions, so to speak, within the organization, in terms of what should be the focus or priority in each. To elaborate further on this, I think I believe strongly that one of the most important thing is to have one data champion within your organization, fairly high up so that, they can have both visibility into what the institutional priorities are, but also to be able to channel and provide the best findings, the most useful findings from data initiatives, to the senior leaders within the institution. And this person’s key role is to really elevate data to insights, and then to tell a story that would then influence decisions and actions at the most important levels. And that really comes from having a robust understanding of the business and being able to talk and engage at the board and the management though.
Angie: That’s such a hard trade-off, isn’t it, that a lot of attractions have to make. If they can only get one hire. Is that sort of an analyst? Do they then bury that in the finance department or marketing department or something similar, ticketing, et cetera? Or is that a leader, that can then show that venue and help connect as you say, the business with the data, to turn data into insights and to really start to prove that right.
Baku: It is certainly a hard decision. But I do again, think that when it comes to to data, the reality is that I think most institutions have very rich data already. And if you look at the market, there are tons and tons of analysts as well as many excellent services and vendors and tools and softwares that allow you to do most anything. But what you need to have is somebody who can help guide prioritization of where to focus and to use these tools in the right way. Otherwise you just drown in the sea of options and choices and resources. So you have to start with the data champion and from the there, you make a more difficult trade off in decisions of the second person you hire, how much focus to put on the technical skills versus business skills.
Angie: Speaking of that champion, a big part of that job, I imagine is really encouraging adoption and usage, rather than sort of simply bringing data to people to give it to them, but rather sort of teaching them how to fish for themselves. For your internal users of data at the Met, how have you gone about that?
Baku: I think it really starts with showing a bit of a proof of concept on what ‘good’ looks like. So, what I try to do is to keep mind of an understanding of who are going to be important stakeholders who will be using data in depth. We can first take on a bite size project and do more of the building process through the centralized data team, but then show, really illustrate the value that we can provide through this. Once there’s a buy-in on this, I think generally speaking, there’s a lot more willingness to learn and adopt and invest in that area from each of those areas. So we would then work with them to focus more on the enablement and training element, as opposed to doing it for them. That’s generally the sequencing of it, but all of this needs to start with being able to show actual values to start out. And that goes back to having a conversation and really good understanding of what the key questions are for each of the areas, what the problems they’re trying to solve, what are the hypothesis they have about the opportunities? And really proving or disproving them through the use of data.
Angie: It’s such sage advice for everybody in this area is  distilling things down into problems and then questions and then hypothesis, or even in some cases, assumptions that people might be making that need to be proven right or wrong.

What about developing data literacy? How have you approached that?
Baku: One of the key things on the data literacy throughout the organization is accessibility and also providing how to read and utilize insights. So to give a tangible example, we’ve created a automated dashboards and reports, which is now actually open, not just to the senior leaders of the organization, if we used to receive such reports, but also to a broader set of people within the institution including mid-level managers and sometimes even all staff. And the key is never to just distribute reports in PDF format and help people take a read, but to accompany with presentation Q and A’s, some sort of sessions to engage them in both describing and having them truly understand the takeaways, but also having them an opportunity to lead to ‘so what’ of findings. That kind of effort takes a lot of time, but it pays dividends in really making people pay more attention to it and having a better understanding and actually having them create more questions and requests that help us make better decisions. So it helps everybody to increase the awareness and understanding and interest curiosity around data.

Angie: It’s funny, isn’t it, it’s sort of that moment when you bring data and insight to a group and you walk away with more actions and questions… it’s actually the successful outcome that you aimed for, rather than everybody sort of smiling and nodding and saying, great, let’s move on. So the work has sort of never done, is it?
Baku: Absolutely. That’s the, that’s the blessing and the curse of the role for sure.
Angie: And what sort of changes have you seen in how your team communicate with each other or collaborate over data or how they make decisions or their team culture? What sort of impacts have you seen with this work?
Baku: I think we’re still certainly in the middle of the journey on this one and by no means  we’re done with this sort of transformation. But one thing I do really enjoy seeing is that people often have conversation at the beginning of the project, or a fairly early on in the project, to discuss how to evaluate and how to think about the success. I think so much of what people used to do was around – here’s a great idea. These are all the reasons to do it. Let’s do it and let’s discuss how to do it. But not so much on what, what does success look like? And numbers are not the only thing for sure. There is a qualitative and quantitative aspect to measuring success, but the fact that people are thinking ahead and thinking about that question as they design and come up with wonderful programs within the museum is very encouraging to me.
Angie: Speaking of that qualitative and quantitative view, I know you recently merged the visitor evaluation function at the Met into your data and analytics team. How do you see those worlds of that traditional qualitative world colliding or complimenting the more quantitative space of analytics and conversely, how does the technology of data analytics disrupt some of those more traditional approaches in that field?
Baku: Qualitative data compliments quantitative data in many ways, they go hand in hand, especially to tell a story. As we have previously talked about the importance of that. Numbers are very cold. There isn’t as much of a sense of feeling and emotions behind it. But once you compliment the quantitative analysis with the qualitative backups, you really put the actual human beings behind data.
And this helps both convey what really we’re finding and what are the things that we can do to improve people’s experience. And technologically speaking, there has been a tremendous progress in many areas, just from the way the world has shifted over the last several years. Couple of examples that really come to the top of my mind. First one is email collection. So COVID has really pushed us into approaching online ticket ticketing more fully. So prior to COVID, most of the people coming to the museum, didn’t buy tickets in advance. They just showed up and bought the tickets through the registers or kiosks. Now the majority of people purchase tickets prior to visiting the museum. And this actually allows us to collect contact information the emails, which allows us to conduct a much better post-visit surveys on follow-ups with our visitors. Turning to on site. The fact that there is a higher and more technologies are supporting on-site surveys, like iPad, the multilanguage surveys and just the sheer quick turnarounds of the findings and survey respondents to be able to derive insights quickly. It has been a game changer on speed to insight. And thinking a little bit about the qualitative data, the world of AI with natural language processing, to be able to analyze thousands, tens of thousands of customer feedbacks through the use of AI to categorize them and quantify even the quality of the comments have certainly allowed us to track what people are  thinking, and saying to us in much more efficient ways.
Angie: Natural language processing. It’s such a good example of how the qualitative and quantitative can come together, even if it is just a first pass, isn’t it? Because previously a lot of those comments would have to be manually coded from scratch by the visitor evaluation teams. So they have to go through and say if they positive or negative or what sort of topics and emotions are coming through and to be able to have that first pass where they can still go and confirm those things and, you know, obviously go and do a lot deeper surveys or processing to get their core takeaways of the meaning that people are giving across. But it does, it does help speed up that process of very manual labor. And focus the time of visitor evaluation on interpretation, as opposed to codifying.

Baku, what sort of obstacles have you had to overcome in executing data strategy?
Baku: I’d say, it was not quite an obstacle, but one thing that did come up as necessity is the, is the need to relentlessly prioritize and focus. There are almost too many things that can be done in the world of data and the museum. And it has really been important to create and maintain and update a multi-year roadmap with a clarity around what we’re focusing on this quarter and next quarter, and so on and so forth, as opposed to taking on all tasks and, and failing at all. This comes back to the resource constraints. If I had a team that’s 10 times as big, obviously we can do more in the shorter time. But just like any cultural institutions, we do have resource limitations. And that really requires us too to know what the focus and prioritization should be.

Angie: And what are some of the biggest challenges you see, if we are looking forward to the future, whether it’s at the Met or more broadly in the whole industry for all cultural institutions, can you leave us with your thoughts on where visitor attractions and working with data are headed?
Baku: Yeah, similarly, I think in general, there’s still a lot of uncertainties when it comes to data as a function, especially in the cultural institution or visitor attractions at large, there probably needs to be a little bit more time and cases being built from in this phase that we’re in with the evolution of the data as a function before we have a shared understanding across the industry of what good looks like. And here are the types of people to hire and here are the types of things people work on. And again, technology is already out there to really unlock value from data that exists, but there really isn’t a sufficient amount of, or availability of the talent and resources who can help bridge the gap between the technology and the cultural institutions, and have a holistic vision for data approach at large. So I think that that is the piece that the next several years will be a time for the industry to really develop and cultivate.
Angie: And it is going to be such a challenge, particularly with the tight labor market we’re in, at the moment as well, for the cultural sector to then compete with the commercial one. It’s going to be a very difficult challenge to address.
[Baku: Certainly agree. Looking at the bright side of it, including myself, I think I think of this as, as a truly, truly a big white space and opportunity for the like-minded people. I am reminded every day, how much value I could be delivering by unlocking values through data. And I’m always feeling constant feeling of need to do more and more and more because I know I can help the Museum make better decisions and actions by working on certain things. It’s just that, you know, there’s a limit to how much one can do in one day. But I do think the same applies for anybody within the institution. And the fact that it is still under developed, or still in the developing phase, really gives opportunity for the people to explore what they can do here in this particular industry, and the areas compared to some of the more well-established industries where the use of data is already codified and cleaned up. And there’s not much more exploration to be made.
Angie: It is fascinating, isn’t it? I think over the past, I think… five years, we have seen data go from experimental and sort of the early innovation of research and then into sort of proof of concepts and pilot projects. And now very much into evolving functions where the focus is on return and deriving value. And really seeing that sort of maturity of the industry, if you like in the data world, it’s an exciting phase to be in.

Baku: Absolutely. 

Angie: Baku, thank you so much for joining us today and for all of that fantastic advice for those people who are going to follow in your footsteps of becoming data leaders themselves and, and for the institutions that will follow in the footsteps of the Met in investing in that function and navigating all of the highs and lows you’ve no doubt been on over the past few years, particularly with the pandemic thrown in there. Really appreciate you joining us today and thank you so much for sharing that journey.
Baku: Thank you very much.

 Ready for more?

Listen to all our other podcasts here:

Download Visitor Excellence Kit

Download Visitor Excellence Kit

As the global leader in big data analytics for visitor attractions, Dexibit takes privacy seriously. Read more about how we treat your privacy.

Download Data Audit Workbook

Download Data Audit Workbook

As the global leader in big data analytics for visitor attractions, Dexibit takes privacy seriously. Read more about how we treat your privacy.

Sign up to our mailing list

Get actionable content sent directly to you. Use these insights to reach visitor management excellence at your attraction.

As the global leader in big data analytics for visitor attractions, Dexibit takes privacy seriously. Read more about how we treat your privacy.