Geolocation data—information on the movement of individuals or commodities sourced from mobile devices and transponders on shipping vessels—has emerged as a valuable dataset in investment firms’ efforts to predict supply and demand and monitor consumer trends.

For example, if you can use individuals’ cell phone signals—which are accurate to within a meter or two—to monitor foot traffic in retail stores and combine that with transaction data from credit card companies, then you can predict a company’s sales and revenues in real time, rather than waiting for company updates, quarterly filings, or annual reports. If you see sales rising, you can buy company stock before it releases its official results, in anticipation of the stock price rising.

Yet the rising cost of compiling and maintaining databases of location-based information and combining it with other types of data is proving too much for some providers.

Within the last month, Gyana, a UK-based startup location data provider turned no-code reporting platform founded by University of Oxford graduate students, closed down, while Denver, Colo.-based mobile location data provider SafeGraph was forced to lay off a quarter of its workforce.

Gyana and SafeGraph aren’t the first location data providers to experience problems. In 2019, New York-based Thasos Group, one of the pioneers of selling location data to financial services firms, had to cut around two-thirds of its staff and its CEO stepped down, according to reports at the time. In March 2020, WatersTechnology reported that the company’s assets had been acquired by Long Island, NY-based Market Service, Inc., which owned various data and business intelligence brands, including AggData—also a provider of location data—and Creditntell. F&D Reports and Arms, which specialize in credit risk reports for retailers and privately held companies.

“Taking over Thasos was a great move for us because we had a diversified client base. For any type of data business, it’s very difficult if you are selling into one market. But we’re a very diversified business selling into many sectors outside financial services,” says AggData managing director Josh Suffin. Since the acquisition, Market Service has integrated the Thasos assets across its brands, built a new dashboard for accessing the data, enhanced Thasos’ original algorithms, and integrated more location data from AggData.

Ironically, Gyana’s no-code platform was designed to help users across industries from private equity to real estate and technology companies to gain insights from mobile data without needing their own data science organization.

“In many ways, Gyana has never been in such a good position. We have a product our users love, and an effective, remote-first culture with minimal bureaucracy and lots of getting sh*t done. Unfortunately, the numbers just don’t add up. Gyana is an expensive product to scale. We don’t have the capital, and thanks to founder dilution and difficult market conditions, a fundraise isn’t feasible,” said co-founder and CEO David Kell, explaining the decision to shut down in a blog post.

SafeGraph’s CEO Auren Hoffman described his company’s decision to cut about 25% of company headcount as a “difficult yet important decision to lower the cash burn,” and conserve cash. “We made this decision even though we had two years of cash in the bank before the cuts. Because in this environment, two years is not enough,” he said in a post on SafeGraph’s website.

But while Suffin says AggData has been able to benefit from location data, he notes that not all providers are experiencing success. “Currently, in the market, we are seeing contraction and companies going away because it’s a costly business to be in, and sales are difficult and—particularly in financial services—can take a long time because of these firms’ compliance requirements,” he says. The key, he says, is to be well-funded, not burn through cash, and not over-hire before having the revenues to support expansion.

“Thasos was doing a very good job with data quality and coverage, they had a great product, and if they’d had more funding, they’d probably still be around today,” he adds.

Mounting costs

Yiannis Tsiounis, CEO of New York-based geolocation data provider Advan Research, says location data has become a much more expensive proposition since Advan was founded in 2015—and providers must be able to offer better service to stand out in a crowded market.

“All things tend to become more expensive, especially because there’s more competition,” he says. “Because there’s more competition, you need more points of interest, and your data must be cleaner. If you want to compete, your data must be of better quality and have greater coverage. So, it’s harder to start from scratch, because if you want to compete, you need to have a solution immediately. We’ve built ours over more than six years. So if you’re starting now, you have to drive faster to catch up to that moving target.”

Thanh-Long Huynh, CEO of Paris-based macroeconomic analysis provider QuantCube, agrees. Beyond simply assembling a comprehensive dataset and augmenting it with related data, there’s a lot of processing that vendors must perform on the data before it’s fit for use by clients.

“It’s taken almost two years for our team to clean everything in a systematic way,” Huynh says of QuantCube’s shipping-related location data and all the information that accompanies it, adding that clients wanting to use the vendor’s data simply don’t have the resources or time to process the data to make it fit for use.

Likewise, he says location data derived from mobile devices is “very powerful” but also very difficult to access and process. “It took us two-and-a-half years before we got the first results. But now that we’re over that entry barrier, I think there are many applications that could be derived from that data,” Hunyh adds.

The effort needed to match that level of data within a short timeframe would normally require a large capital investment of anywhere between $20 million and $200 million, Tsiounis estimates, whereas Advan was able to build its full offering over time.

“We never relied on investor capital to survive,” he says. “We’ve been cashflow positive for around five years now, we keep growing, we have long-term clients, and we spend within our means.”

But keeping on target is hard when others are constantly moving the goalposts. For example, as demand increases, the underlying data sources realize their data has value, so they charge more for it, Tsiounis says.

In addition, location data on its own has little value: it’s when that data is combined with other information that it becomes valuable. Location data in stores doesn’t tell you whether customers are buying goods, or how much they’re spending: Point-of-sale transactions data does that. Likewise, the location of a ship on its own says nothing. But knowing what kind of vessel it is, what it carries, and using shipping manifest data to understand what it’s carrying, and where it’s sailing to and from, creates a much richer picture—and one that can be used to make assumptions about oil prices, based on supply and demand.

In QuantCube’s case, the vendor collects more than 14 billion data points that encompass much more than just location data, many of which augment that data and enable users to draw more accurate conclusions—for example, in addition to tracking 80,000 vessels and updating their position every 15 seconds, the vendor also captures atmospheric data from satellite photographs, and monitors social media in different languages most relevant to specific commodities. For instance, social media posts in Arabic contain five times the number of oil-related posts as English-language social media, Huynh says.

Those extra datasets come with a price tag—not to mention the cost of a team of data scientists to marry the datasets and derive insight that delivers value for clients. And that’s also a moving target: As vendors and their clients identify new use cases for the data, vendors need to license new datasets to support those uses. For example, using mobile device data to monitor foot traffic at a residential building can be a real-time indicator of occupancy rates, which can serve as an input to pricing commercial mortgage-backed securities more accurately.

Another use could be using data related to the mobile devices—within the confines of the General Data Protection Regulation (GDPR)—such as whether that device’s owner (an individual or a corporation) has paid their monthly cell phone bill or not, to use as an input to credit risk assessments, Huynh says. Or, by analyzing the mobility of company-issued mobile phones, an investor could expect to see higher or lower revenues at that company based on how mobile its staff are.

And as more datasets emerge, these in turn create the potential for new use cases. “There are certainly lots of alternative data points being created by the internet of things/wearables, app installs and geolocation, but they are not as widely used as other ‘alt data’ sets,” says David Easthope, head of fintech in research firm Coalition Greenwich’s market structure and technology team. And while noting that “geolocation or footfall is more of a specialized dataset used fairly selectively by certain hedge funds,” he also adds that geolocation data often appears on the future roadmaps of funds not currently using it.

In the timeframe of those future roadmaps—possibly within three to five years—QuantCube will be able to estimate revenues and financial statements for companies in real time using combinations of alternative data, Huynh says.

Context is king

However, location data needs to be consumed in conjunction with those other datasets to have value. “Location data is only one part of a whole lot of analytics we need to analyze macroeconomic trends or international trade. It’s extremely powerful, but not just by itself,” Huynh says, warning that the market will not be large enough to support a multitude of pure location data providers, and that vendors will need to diversify to provide a full data offering, of which location data is one component.

And key to getting the most out of that data component is being able to understand properly how location data can be used. Once a firm or vendor fully understands how to view information provided about and by a location, they can grasp how to develop any number of potential analyses and use cases.

“This data is so massive in its scope and also its opportunities,” says Gregg Katz, head of product and vice president of real estate strategy at AggData’s sister brand Creditntell. “Having the data is one thing. Understanding it and knowing how to educate potential clients is something else. And this data can go wrong very quickly—by which I mean you can get bad results—if you don’t have the right context around it.”