Maxim Matias
Mar 31 · 10 min read

Understanding the potential of human mobility from a conversation with Thomas Walle, CEO & Founder of Unacast (previously part of the founding team at Tidal)

Found on Shutterstock under Time Square Night

Personalization has been a subject of debate for decades, since the time when it meant no more than including someone’s personal info in an email. In the last decade, the sophistication of targeting technology has undoubtedly reached new horizons, yet there remains a sense that the potential is still to be captured. The defining global tech companies are built on “online” graph data representing 30% of human activity, but 70% of human activity happens in the physical world. A new paradigm shift has emerged, waiting to be conquered, fueled by human mobility data.

Unacast is a company that defines human mobility as an understanding of how people interact with places over time. As its founder and CEO, Thomas Walle says: “At Unacast, we understand how people move around the physical space, places they go to, where they travel to, where they live, where they work and so on. Once analyzed, we then provide these interesting insights to companies - so they can make better decisions and better products”.

Relevant examples and on-going case studies are:

Mike Reiner, Venture Partner at OpenOcean interviews Thomas Walle, CEO of Unacast:

Mike: What got you into location data in the first place?

Thomas: It might not seem like the obvious starting point, but me and my co-founder Kjartan Slette, before Unacast, we were a part of the founding team of a music streaming service named Tidal, that was acquired by Jay-Z. As you can imagine, we knew every single song that people listened to, but what we really wanted to figure out was where and when people go to concerts. This was back in 2013 — we understood the online space, what people clicked, what they purchased and what they listened to, but we had no idea about the physical movement of the people. Once we dug a little deeper, we realized that many companies, spreading around various industries would heavily benefit from such information.

Mike: What data sources are available to you to understand the physical movement of people?

Thomas: As you can imagine, this has been a long journey for us as a company, as we’ve had to figure out which data sources would give us a clearer understanding of how people move around in the physical space. We use sources such as:

  • GPS data that we collect from apps and publishers
  • Cell tower data
  • Wi-Fi data (logins into public networks at shopping etc.)
  • And soon purchase receipts from shops etc.

In order to get a clearer image, you really need to look at a lot of data sources and find a correlation between various factors.

Mike: Which one is the most relevant for you?

Thomas: GPS data is probably the most common in the marketplace, as there is a lot of it and most applications are now GPS enabled — primarily for studying your whereabouts. But, it also gives some challenges, as there is a lot of that data available and you really need to figure out the quality aspect of it. It is crucial to enrich your data, by using many available data sources and be able to formulate a clear image of the physical movement of a customer.

Mike: How do you deal with data preparation?

Thomas: This part of our business is very exciting, as you get to observe how this type of data has been created. If you look at GPS data, it’s the combination of a tiny GPS chip in our smartphones and then connected to 31 satellites orbiting the globe at 12 thousand miles above ground. Think about those numbers for a bit, the margin of error between those is massive. This is a very challenging aspect of GPS data. To get a precise understanding of how people move around in the physical space, you really need to be able to source the right type of location data. You need to be able to interpret the data correctly in order to build this knowledge.

This is where a lot of companies are failing to recognize how hard it is to deal with GPS data compared to other more traditional data sources. And it is crucial for companies to understand the importance and potential impact of applying such technology.

Mike: There are many retail analytics companies out there right now that are using location data inside of their buildings for different use cases. What is your intake on that?

Thomas: What we see in that space is something that companies are still trying to solve to this day. However, I believe that the use case of inside stores is being solved and, in many cases, are even quite sophisticated. The retail analytics space has definitely matured and grown, as there are tons of technologies that effectively solve these type of problems, either via camera, sensors and other devices that are capable of telling something about people’s behavior in that particular store. But what many of them are missing is what happens when the customer leaves the store or where they come from?

I believe this is one of the biggest challenges for physical retailers today, simply because the predominance of the dataset they have is not used to the full potential. Common questions that need to be asked, are:

  • What’s happening in the store?
  • Who’s buying in my store?
  • What about all the people who are strolling through the store without buying anything?
  • Where did the customer come from? Where is the customer going afterward?
  • Did they come from competitors?
  • and so on…

This is where we see a lot of very exciting use cases now with our clients and how these in-store retail analytics companies are leveraging location data to get a better understanding of their customer.

Mike: What are your lessons learned in terms of data-sourcing?

Thomas: Ha! Data sourcing is super tricky and as I mentioned with GPS data, in the early days of Unacast — we really had to spend a lot of time and energy to understand what a good location data signal is, for example — How can we, in a good way, understand as much of the user as possible throughout the day. We’ve approached it from multiple angles over the last couple of years — the number of data sources that we’ve examined is huge and it really is like finding a diamond in the dirt. Primarily due to vast amounts of data out there, but very minimal amounts of quality location data.

Another challenge that has appeared is when monitoring a customer. We’ve only managed to see various users once per week, or once per month. In order for us to get a better understanding of the user’s behavior, we need to see the actions of the customer at least multiple times per week, preferably multiple times per day. So, to summarise the learnings and challenges, it’s pretty much understanding the right data and finding quality data within the vast amount of data out there.

Mike: Quality over quantity. Could you share any best practices in terms of how much qualitative data you gather?

Thomas: Frankly speaking, it’s roughly 1% of all the data gathered and that is ready to go into our core products. This shows you that most of the data is not sophisticated and detailed enough for us to understand human mobility, yet.

Mike: If you’re saying that 1% of the data is qualitative. Are you guys trying to get this number up, or are you planning on gathering more data and figuring out for yourself and translate the data in a more meaningful way in the future?

Thomas: This is where we see a transition in the human mobility space, as for the last couple of years it’s been very much about the volume game. The more data, the better. This has advanced, as we can see from our use cases and the level of sophistication from our clients. Products are becoming more sophisticated and the overall trend is shifting to ensure that we have quality data and try to solve the current pain points in human mobility.

Mike: When speaking about data sourcing, is there a difference in data collection from different regions? For example, comparing the US to Asia.

Thomas: There is a big difference. The US market is more sophisticated. The US is leading this industry and they have done so for the last 4–5 years. We also see Europe is increasing in maturity and accepting the importance of this field. Regarding Asia, it seems they are a bit behind, even if that region has a mobile-first mindset. We know that there are tons of apps with very sophisticated location data strategies such as WeChat for instance. We are expecting Asia to go ahead of Europe in the next coming years. As many of you might know, GDPR came into force in Europe on May 25th in 2018, and many companies fled Europe, as it didn’t allow these companies to gather enough data to successfully interpret results in mobility and hence not become GDPR compliant.

Mike: How do you ensure your clients use the data for the right use cases?

Thomas: That is something we are internally discussing for more than 12 months now and I personally believe that if you want to increase innovation by the use of human mobility data, the client needs to understand what their data consists of. This is where we as a company are different from many others in the industry. We don’t sell a dashboard with data reports, but we sell a tailored dataset to our clients to enable companies to make better decisions and therefore increase the quality of their products. This also reflects deeply into our Norwegian mentality, so to speak, and this is telling our clients exactly what we have done with the data. An example here could be that when taking a specific customer user case, and this customer has spent about 25 minutes in your store — we also tell the story of how we arrived at this conclusion, how many data points we used and show through how many partners this confirmation went through that one customer was at this specific place. A customer is about 70% maneuvering in the physical world and our clients and we will only succeed if we fully get an understanding of the physical mobility data.

Mike: You’ve recently published an article about the location data ecosystem and how it’s split into 3 key layers. Could you elaborate on that?

Thomas: The mobility market is still a fairly new market, but we are starting to see that different companies are taking different positions.

The 3 layer pyramid below should help to visually understand this.

Pyramid created by DataSeries from the content of Thomas Walle

Starting from the bottom:

Tier 3 — Here we usually find compressed companies with a size of 5–10 people who display commercial experience and focus very much on licensing and selling data. This is usually done by the compiling of information from databases with the intent to prepare combined datasets for data processing.

Tier 2 — This tier consists of human mobility companies (Unacast fits in here). The mechanics here are all about understanding raw location signals of data and get an insight about its meaning. However, these companies don’t necessarily build end products or fully-fledged solutions. This is where the T1 companies come into play.

Tier 1 — Companies are leveraging location data for their core applications. That could, for instance, be a real-estate company studying the foot traffic near a neighborhood to distinguish whether it’s becoming a popular location or not. This is also used for use cases such as how many people are commuting from the suburbs into the city every morning if one main-road is shut down for road maintenance. Here it would be important to see how many cars will be redirected to the surrounding streets and how that could impact traffic.

Mike: How do you guys source data today vs. in the beginning?

Thomas: This has been an evolving journey. We had a mission, we wanted to understand human mobility. So, we started working beacons and even though many were optimistic, beacons didn’t take off as people expected to. It’s hardware, it takes time to scale. We, therefore, transitioned to GPS data as a core data source, which can solve the same challenge. It is difficult to place a mark on when our strategy has changed, as it’s an ever-evolving process. I wouldn’t be surprised if our team says tomorrow: “Guys, we need to change how we source data again”, “There are some new ways that we need to look at the data, so we can make the best decisions and get the best insights about human mobility”.

Mike: This ties nicely to the last question, which is, how do you see the future for both the human mobility data and your business?

Thomas: I truly believe that location data is going to be a vital touch, a vital part of any decision-making for both human mobility and the products that we interact with. Keep in mind, that up to this point, we’ve only been able to understand online data, and for the most part, it only captures 30% of our awake time. Finally, there is also a way to leverage the remaining 70% and I believe this is going to infuse so many companies and verticals. More and more industries now start to realize the importance of human mobility data and the decision-making force it brings along.

Coming closer to understanding the rich insights of human mobility, can transform the way cities and transportation services understand their citizens and customers and empower them to adapt to the needs of a growing population efficiently. By being able to explore all fields possible, personalization will reach new heights to customers and clients and eventually accelerate the overall decision-making processes.

DataSeries

A network of data thought leaders, sharing lessons learned, in preparation for the future 🚀

Maxim Matias

Written by

Venture Associate @openocean ; building a data community at @dataseries ; MSc @imperialcollege ; Contact at: https://www.linkedin.com/in/maxim-matias-ba533666/

DataSeries

A network of data thought leaders, sharing lessons learned, in preparation for the future 🚀

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade