The Data Product Mindset: Peter Bjørn Larsen

Data beyond tables: How to design and fund powerful data products

This latest episode is your blueprint for creating powerful, in-demand data products.

Harbr founder Anthony Cosgrove talks with smart city expert Peter Bjørn Larsen about creating multi-modal data products with user-centric design and ROI at the core. Join their conversation as they go deep on data product management:

• Moving beyond tables: Discover how to design multi-modal data products that include APIs, dashboards, and apps to cater to different user needs.
• Funding your data dreams: Learn how to identify a wide range of use cases for your product, opening up diverse funding opportunities.
• Template magic: Maximize your return on investment by using the same data product template across different contexts.
• Decoding user desires: Uncover the secret to understanding what users truly want. (Hint: it's not just the data, it's the actionable insights.)

Transcript

Anthony: Hi everyone, and welcome to episode six of Data Product Mindset. I'm Anthony Cosgrove, the co-founder of Harbr, and today I'm joined by Peter Bjørn Larsen, who is the founder of Smart City Insights. Peter, thank you for joining us. How are you doing today?

Peter: You're welcome, and thank you. I'm doing extremely well.

Anthony: Amazing. Peter's joining us today from Copenhagen, which is very exciting. Obviously I know quite a bit about Smart City Insights and what you're doing there, but for everyone else, can you tell us a bit more about it? What are you up to?

Peter: For the past 15 years, I've been working with urban development, data and technology, space data exchanges, and other kinds of data products relevant for cities and their stakeholders. Right now we have a big project here in Denmark in Copenhagen called Urban Insights Data Lab, where we're creating different types of data products for various stakeholders in urban development with public and private sector data sources.

I just got back from Japan, where we looked at some of our results to test them in a very different setting in Tokyo, with a great response. We're also talking to people in the UK and other places about it. We're doing a lot here and trying to figure out how we can replicate that and use it elsewhere.

Anthony: Amazing. You mentioned data products and clearly you're thinking about reusability of similar patterns in different jurisdictions, different urban contexts. Tell us how you think about data products. What are they? What are they not? How do you approach that? I know you've got some of the most interesting examples I've heard of data products, so it'd be great to hear about ones you've built or come across or used.

Peter: Let me go back in time about 13-14 years ago. I worked with the city of Copenhagen when I was at Hitachi, and the city wanted information about what was happening there as they were developing a new climate strategy. There was a suggestion to create a city data exchange, which became its name. We discussed this with the city of Copenhagen and different data stakeholders who had interesting data about the city. In 2012, we ran a competition to create a data exchange where we could actually buy and sell data.

I eventually got hired to be the director of this. It was a data exchange for Copenhagen where we had many different stakeholders involved. This was also when GDPR regulation came out. We learned a lot from that, both in Denmark and across the UK, US, Australia, and all over the world. We had discussions about how to manage and use this data.

The project we're doing now, Urban Insights Data Lab, really represents the learnings from the past 10 years, not just from the city data exchange but from all these types of data product projects. We see telcos, for example, have different kinds of products they sell to retailers, cultural institutions, and cities. We looked at telcos, financial transaction companies, credit card companies, but also the offerings from different public sector open data portals, to see how that actually fits into the demand we see in urban development.

The products are very different. It could be a dashboard. It can be an API you can get from an open data portal. It could be different kinds of flexible and non-flexible products. There's really a lot of different types of data products and data collaboratives, like what we saw during COVID where companies came together to solve problems by combining different data sources. That's what we've done here in the Urban Insights Data Lab as well — identifying what data sources we need to solve problems in urban development, and those are the data products we use.

Anthony: Amazing. And how do you handle these public and private sources of data? How do you tease out the use cases? Are you constantly interacting with those same organizations? Does it go down to individual consumers and their particular needs?

Peter: We try to look at what makes up the urban development ecosystem. Of course you have the city authorities who have responsibility for urban development. Then we have the developers who are building and need to understand different urban areas. We have architects, different kinds of cultural institutions, mobility companies, supermarkets, shopping centers, retailers.

We brought all of them together because they all play a role in an urban area. Then we worked with them individually to understand their needs, not talking about data initially, just asking what information they need. The architects tell us their needs, the developers tell us theirs, and the shopping centers share theirs. Then we put this demand for insights together and use my colleagues' knowledge to identify where we can find this information and how it matches what they're asking for.

Peter: It's been quite a journey because there have been some great data products, but they didn't exactly fit the needs, which is one of the reasons why they didn't really use that data. What we've spent the last two years on is trying to figure out how we can convert these existing data products from different data sources into something more useful.

Anthony: Can you talk a bit more about the urban development journey, from people having some demand, finding data that didn't quite stick, to actually going further? What does that look like?

Peter: One of the data sources we use is a telco in Denmark, which also operates in the UK and many other countries. They know about the movement of people, which is interesting in all urban areas, but their data products were limited. You could zoom into an urban area, an event, a sporting event, a cultural institution, or a shopping center, and they could tell who is there, who's coming in, the geography, age differences, gender and so on.

But when we talked to the cities, architects, and developers, they said they wanted to know about the citizens. They wanted to know about people within an urban area—how long do they stay, when do they leave, where do they go? Do they go to an outside shopping center or the city center? What are the different age groups that leave? When you do planning for an urban area, you want to know what's keeping people there.

They didn't have that product before. For tourism, if there's a conference somewhere with 3-4,000 delegates, it's not so interesting to just know they're there. Wonderful Copenhagen was one of our use cases—they try to spread out the tourists, and we can actually measure if their interventions are working by looking at the data.

This wasn't a data product before. We explored how architects could use this, how shopping centers could use it, how the city could use it, and developers too. It was only possible to change the telco's data product because we made a market for them, showing that five or six different stakeholders within urban development would be interested in their data if they changed it.

Anthony: Wow. So really getting down to not just a use case, but actually specific personas, what they need, and very precise requirements, like a postcode as a field is unhelpful for that use case.

Peter: Exactly. If you look at it as a suburb to some city, somebody might be interested in the whole suburb. A developer might say, "We've built apartments in one part of that area, we only want to know about residents in this area." The shopping center might say, "We just want to know about the shopping center." That's an example of three different geographical needs. Or if you made an intervention, that's what you want to measure. The flexibility in that is extremely important.

Anthony: You mentioned terms I haven't heard anyone use before, like a flexible data product. We've spoken about this on some of the previous podcasts—how people broadly want the same thing, potentially in the same format or delivery mechanism, like a chart or a graph, but then there's still some level of flexibility required to hone in and be more precise. How do you manage that? Because that introduces cost, right? How do you think about and approach it?

Peter: The requirements are different in terms of expertise in using data. If you're a large city with a data department, they might say, "We have our own programs. I just need an API. I just need to get the data in and we'll handle it from here. I don't need your dashboard or graphs." Others need a dashboard and graphs, and the ability to click around. The market demand differs, but you need to build for probably the dashboard and graphs to have that ready. Then a side product is just the API, which might be cheaper. So we can offer both raw data, graphs, and dashboards. Those are the three things we see that people want.

Anthony: So almost approaching it expecting to have like a multi-modal data product where the base data is the same but the consumption mechanism is different. And just having different consumption mechanisms provides flexibility, but then some of those might have charts with sliders and views and other things.

Peter: Exactly. And then to make it a product — a lot of these data sources have been asking you to provide information about a certain area. Of course they can, but it's what I call handheld. Their data scientists need to do a lot of different types of work where it's not still a product, and it makes it very expensive. They do that because they think it might be a one-off or two-off.

This is why we try to say it's important to understand that urban development is a market and not just cities. That has meant these different data sources have created data packages, not just data packages, but products. There's been a cost, but we also had some funding to help them in this project. Now these different stakeholders can get the data cheaper, where the data sources still get revenue and profit, but the costs are very different because it's a product.

Anthony: So a bit more upfront investment, but eventually sort of pays its own way in terms of scalability and ease of use.

Peter: Exactly. And then there's different ways to finance and fund that. In Denmark, and I know in the UK as well, there's national funding available for these kinds of things. We looked at everything from small town centers, which is a big thing here in Denmark right now—they need a strategy but don't have any data. So there's funding to develop these products that can help them understand the town centers and continue to evaluate them. We know the data sources now, but there's still work on flexibility and these kinds of things.

We try to find themes of national importance that have funding available within urban development. Then we see how we can combine the development of data products with something that's on the radar for these different funding sources.

Anthony: So really finding the topical problems that people are solving now.

Peter: Of course, climate and net zero are big ones. I speak with UK stakeholders about what information we need and where we get it from to measure net zero for investors, cities, and other stakeholders. How do you put that package together so it's replicable from city to city? This is important—if you do it for one place, you should be able to repeat it in a town or larger city.

What we set in this project is that it has to be scalable. This is the only way we can work with the data sources and the funding. We're testing it here in Copenhagen, but it's actually not the area in Copenhagen that's interesting. We looked at a small town in Copenhagen, but you can copy it and use it in other Danish cities and urban areas. The same thing applies in the UK or Australia or even in Tokyo — you can use it with local data sources and the same model.

Anthony: So essentially the same schema to a table and then the same pattern of outputs.

Peter: Yes, and we're using data sources that you can find elsewhere. There's a lot of data available from university projects or an app, or from questionnaires somewhere, but it's not repeatable. If you want to work with urban development, we're not talking about something that happens within a year, but quite a few years. I've seen over the years that if you really want to create a data product, you need to make sure it's available for many years in the future. You can never promise anything, but they're just not going to invest in it if you don't get the data.

Anthony: That's such a great point. I see debates around whether a data product needs to have more than one user. Hypothetically, it doesn't have to—it could just be a really high-value product. But the reality is that most people need to find multiple avenues to fund it and get it going, then pay for it on an ongoing basis. In some cases, there's a huge advantage to doing that because you really want the largest addressable market anyway, and you can get the ROI and profit and margin and all good things. It's so interesting listening to you—it feels like you've got this idea of creating a product and finding as many different use cases for that product, but then also thinking about that same product in a completely different urban area, like a different complex, and just repeating those patterns to see how they scale. So you've got growth of use case locally, but then the ability to repeat and reuse the same pattern in multiple places, which feels like a really good way of securing funding and giving people what they need.

Peter: Exactly. One of the interesting parts is that they know what they want, but they still don't know how to use it. So there's some training involved because this is new information. Typically when we talk to stakeholders who work with urban development, they always think about the data they know exists—some socioeconomic data, maybe a bit about pollution, maybe a camera. You need to stop them thinking like that because they only come up with suggestions of what they already know. We tell them to just tell us what they want and make a wish list. In the perfect world, what do they want? We might not give them a hundred percent, but we might go from 40 percent to 80 percent.

We've spent the past two years educating big companies, developers, and cities to just talk freely about what information they want about an urban area, and then it's our job because we know where to find it. That's the data broker role. Then we get the data and they need to learn how to use it, so it goes forward and back. This interaction between the end user and the different data package owners has been really important.

Anthony: Amazing. I usually ask about experiences and lessons learned, but I feel like we've covered about five or six of those in just the last 15 minutes—from how to think about the market to the importance of user interaction, building flexible data products, personas, and different mechanisms for servicing those personas. You're in a really unique place, building these data products on this long journey from marketplace to creating multimodal products. Where do you think it's heading? What's next?

Peter: There are hopes, and then there's the way I think it's going. In our current project, we have individual data packages from different types of companies and the public sector, but individual data products don't solve the problem of an airport or an urban area. It's a combination of these data products that need to be combined, or at least described in metadata within what I call a data collaborative.

When potential users ask questions, we need this kind of data broker function on a platform or a person that can help them put together the different data products. That's definitely a big wish and need to get the market to actually use the data more. Of course, there are digital twins and these things, but I really think we need to put data that's relevant and what they're asking for into that, and then they will start to work as well. But there's still a while to go with digital twins, I think.

Anthony: Cool stuff. I could literally sit here and chat for another two hours, but we'll have to end it there. Thank you so much for joining us. If anyone listening is interested in learning more about urban development or what Pete's up to, you can check him out on LinkedIn — Peter Bjorn Larsen — and he's regularly posting. And if you're building data products or a data marketplace, check out harbrdata.com. Thank you for listening.