Geolocation data
The Augvest discussion held on 22 February 2018 featured four vendors of geolocation data:
Cuebiq's offerings range from visitation data delivered via AWS S3 or API, to footfall data delivered in Excel, to a SaaS platform to further help visualize trends. Cuebiq also captures handset activation and market share by manufacturer, model, and wireless carrier.
As of February 2018 Cuebiq has 26 months of history and access to 2 out of 5 smartphones worldwide, including 72m in the US and 600m in China. In the US they also have a large "power user panel" that represents devices that have appeared with high frequency throughout the history of their data.
Close to 600 apps are using the company’s SDK - mostly local media, weather, and travel apps. The company currently has 34m daily active users and is on track to get to 50m daily actives.
In addition to offering geolocation data on more than 50m phones, SafeGraph maintains a database of 15m points of interest (POI) and records 300m visits per month to about 1,000 brands.
Samples of research done with Safegraph data can be found on their blog.
Previously founded as an app to prevent drunk dialing, X-Mode has now developed a mobile location SDK used by 300+ apps ranging from dating to edgy weather apps.
Provided users opt-in, the X-Mode SDK is always on. X-Mode collects an average of 100 points per user per day, 9.5m / 30m daily / monthly active users, and can pin-point 90% of locations to within 20 meters.
The panel was moderated by System2, which uses location data to forecast store closures, measure macroeconomic trends, and predict consumer behavior. Lots of work goes into normalising this data - details will be presented in a workshop to be run in partnership with GLG; contact us for details.
General observations
Traditionally scale location data has come from cell towers - but that data tends to be far less precise than GPS or wifi.
Another source funds can turn to is the ad marketplace, but less than 10% of that data is high-quality (source: Safegraph).
Geolocation firms compete first and foremost on how they can help apps scale. This naturally leads to specialisation; though there is some overlap, each of our panelists generally works with different kinds of apps.
Revenues from selling data to investment firms are just icing on the cake. This parallels what Augvest has seen with other alternative data sources - corporates tend to have much larger data budgets.
Even if location data is accurate, analytics are complicated by several common issues:
Difficulty of tracking individuals across devices. Common solution is to look at where individuals work and where they live. There are also dedicated solutions like Drawbridge.
Each data vendor presents a biased sample of the population. Combining different location datasets helps address this but also requires a lot of work to be invested into normalising and deduping data.
Normalisation is also complicated by rapid growth in panel sizes.
There is little data going back more than two years - year-on-year changes often only beginning to become available.
Regulations continue to be a source of concern:
General Data Protection Regulation (GDPR) in European requires users' consent not just to having their data collected but also to how that data would be used. Similar regulations are under consideration in California.
It can be helpful to have independent intermediaries do more sensitive work before passing aggregated data to funds