Location Data made simpler
Location-Based Services, and Applications using Location Data, have a bit of a history. And from a business perspective, not a particularly glorious one at that. For some time, experts have been boldly predicting that the next year would be the Year of Location, or Year of Location-Based Services. This has led to a turbulent journey these past 20 years, as new applications have generated new waves of optimism and forecasts of multi-billion $ markets — and then simply disappeared. Initial booms were fueled by consumer services that used location, but in recent years the emergence of analytics based on location has promised further ways to create new business.
In this article we look at Location Data and try to give some help and insights so that you can assess the quality and legality of any Location Data. It will focus on location data from mobile devices and smartphones in particular, though can be applied to other things and their associated applications. I’ll put my own perspectives forward, hopefully supported by your experiences working in this field, or as a consumer of Location-Based Services.
Although Location Data has not fully delivered against business expectations, advances in areas such as Smartphones, Visualization Tools and Artificial Intelligence, may give it (by my counting) the 6th chance to prove itself. Location Data comes alive and creates insights when it is put to use in Applications and Services, standard examples of which would include:
Location-Based Advertising — Receiving offers from businesses that are nearby.
Search and ‘Find my Nearest’ Services — Relative to where I am. Shows me on a map.
Navigation and Traffic Services — Showing you the way to reach places in the most efficient manner.
Asset and People Tracking — To keep them secure and safe by knowing where they are.
Emergency Caller Location & Car Accident Location — To get help to you quicker.
Segment and Audience Creation — You are, where you are. Categorization by the places you are located in.
Dating — Always more powerful when people are located nearby, my friends tell me.
Mobility Analysis — How groups of people move, and traffic patterns and levels, are useful for applications such as: Retail/Store Planning, Public Transport optimization, Financial estimates using footfall, Smart City Operation, and more.
It is said that your Location gives the ultimate statement of intent, involving as it does the real world and some physical effort, as opposed to a few seconds of surfing the internet on a sofa. Perhaps because of this, location data carries a certain amount of weight and value. But not all location data is created equal, so we will look at the various types of Location Data powering these Services, focusing mainly on the 2 most important and contentious dimensions of Accuracy and Lawfulness.
2. Structuring Location Data
There are several key dimensions and properties that can be associated with any set of location data. Aside from obvious stuff like how much data there is, and what territory it covers, some of the more fundamental metrics can be expressed generically in a format with 5 basic components.
As an example, consider that:
- A person is located at map coordinates (48.137247, 11.575698), or at Marienplatz in Munich.
- By the standard hybrid calculation method used by smartphones, with a high GPS component because of a good sky view, giving up to 10m of error.
- At 14:32 on Saturday 8th February 2020
- They are identified using the Advertising ID that the Google Maps Application uses to identify them.
- They allowed the Google Maps Application on their iPhone to locate them by selecting the ‘While using the App’ option in the iPhone Privacy settings under ‘Location Services’.
This can be considered as a single item of Location Data with 5 associated properties.
3. Accuracy — From vague to sharp
This is the most obvious metric for location data and one that creates the most debate, much of it being subjective and anecdotal. Consider the following levels of accuracy and examples of technologies that operate at these levels.
The underlying technologies used to generate the location data define the accuracy and therefore what Applications and Services can be supported. can be done with the resulting data. They will often also define the quantity of data available, as well as giving a hint as to how such location data can be legally obtained.
4. Common misunderstandings around Mobile Location Data
When it comes to the accuracy of Location Data there is a 20-year catalogue of claims, distortions, selective testing, omissions, exaggerations, and marketing. This makes distinguishing fact from fiction difficult, so here are my 5 favourite myths around location.
The Location Data Phone Apps use is ‘GPS Data’ — This is very wishful thinking. Phones typically employ a hybrid technique taking all received radio signals (GPS, Cellar, WiFi, etc.) and calculating a position. There may be a strong GPS component if there is a good view of the satellites, but indoors or inside a shopping centre or mall (a common use case) you will often not get ‘GPS Data’.
We use ‘Triangulation’ to get more accurate Cellular Locations — In nearly all countries in the world this is highly unlikely to be true. Theoretically there are various ‘triangulation’ or multi-cell calculation methods available, based on received signal strength or timing information from 3 or more cells, but they have generally not been implemented. This is due to both the high costs versus the business benefits from a relatively marginal improvement in accuracy, and the growth and ubiquity of GPS and WiFi.
Cellular Networks locate to 100m accuracy — Cells in a mobile network typically vary in size from 100m to many km. Whilst busy city centres can occasionally return 100m accuracy, these form only a very small % area of a country. As you leave these areas the accuracy degrades, quickly sliding out to 500m (a likely nationwide average), and >1000m as we reach the suburbs. You usually see the best accuracy figures quoted (e.g. up to 100m) but they are not typical, and nationwide or citywide averages are much less accurate.
Many new location-based applications will be enabled as we introduce 5G / 4G / 3G — Unless laws of both physics and base station economics have changed, this is not really the case. There can be marginal improvements as the Mobile Generations and technologies change, but the location accuracy is still a fundamental function of cell sizes and cell densities, which have not changed radically as mobile technologies have evolved over 30 years.
Don’t worry, the Location Data is completely Anonymous — Watch out for this word when discussing Location Data especially. It does not take much additional data or processing to turn Location Data into a specific person. Generally, most people can be uniquely identified by 2 or 3 locations, such as their home and work.
5. Sources of Location Data
Mobile Network Operators
Location Data is a by-product of running a mobile network and further Data can be extracted associated with network and user events. The challenges concern the relatively small numbers of consented subscribers, and the complexities of thoroughly anonymizing huge quantities of Location Data. Furthermore, mobile operators are generally regulated via licenses and telecommunications law, bringing extra sensitivities. Technically mobile operators could extract hundreds of (cell-level) locations per subscriber per day, combine them with CRM data via the phone number (MSISDN), and then know to a detailed level how groups of people move throughout the days.
In theory the owners of stores, events, shopping centres and similar constrained places could track their visitors. Gaining user consent for any profiling and re-targeting is not straightforward. Premises WiFi Networks can sniff and identify nearby devices, but the MAC Address identifier is not commonly used and MAC randomisation makes the building of such profiles difficult. Going down the anonymisation route, provides more limited data such as general visitor traffic levels.
Aggregated Bid-stream Data from Advertising
When there is a space for an Ad in a mobile browsing session or App, the Location Data and unique identity of a user can be broadcast to many players in the advertising ecosystem, who then bid to target this user. The winning bidder delivers an Ad to that space/user. However, all the losing bidders have acquired a location and an ID, which they could use for profiling. This system is known as Real-Time Bidding (RTB) and is the core of Programmatic Advertising. This mechanism has come under some scrutiny, as the broadcast and processing of users Location Data by so many parties they have no relation to, appears to bypass GDPR consent considerations. Use such Location Data at your own risk.
Many Apps ask to use your location and the respective operating systems (Android, iOS) allow users further control of this data. App Terms & Conditions should describe the personal data they use, and for what purposes, and users then need to consent to these. Complications occur when data is passed to other parties who operate B2B so have no relationship to these consumers. Location Data can be extracted when the App is being used (foreground), and when the App is not being used (background). Foreground locations are infrequent and unpredictable, being dependent on when specific Apps are used. Background location, usually on a periodic basis, has been made more difficult as first iOS and then Android, have restricted its collection. It can be done, but the App needs a solid reason to collect location data, so typically it’s Apps such as Weather, Dating, Local Search, Public Transport, Fitness, Mobility, some Games, and not that many more.
Finding good quality and lawfully obtained location data is not always straightforward.
Smartphone Apps are an obvious source, though the Location Data may not be fully GDPR-compliant if your processing is not essential for the App’s specific function.
There are clear ways to design Apps to be GDPR compliant, and to make Location Data available, both personalised and anonymised.
·Beyond Apps there are other sources and owners of Location Data that can be approached, and their Location Data assessed.
Anonymisation of Location Data reduces the requirements on subscriber consent and can increase the amount of data available. However, the data value may reduce as other attributes cannot be invoked, nor can the user be re-addressed.
It may not be easy, but it is feasible to use Location Data to improve targeting, planning and decision-making for businesses across different industries. It requires some Location Data expertise.
1. GPS is used in the text due to laziness but could mean any Global Navigation Satellite System (GNSS) system, including GPS, GLONASS, Galileo, BeiDou.
2. The article only focusses on GNSS, Cellular and WiFi location technologies, as these are felt to be the most important and widespread since the start of mobile location until today. We could certainly use a new location technology after all these years.
These are 100% my views and in no way represent any company I am, or have been, associated with.
About the Author
I’ve spent the last 20 years working with Location-Based Services and Location Data in one form or another. This has been in Product Management roles for companies at various points in the Ecosystem, including: Consumer Services Provider, B2B Advertising Solution Vendor, Location Technology Vendor, Navigation Solution Integrator, Map Builder. I have found that a common understanding of Location Data, and a means to assess it, is not always easy to find — so I wrote this.
- A new report by the government-funded Norwegian Consumer Council (Forbrukerrådet) on the use of phone data (including Location Data), by the Advertising industry. https://fil.forbrukerradet.no/wp-content/uploads/2020/01/2020-01-14-out-of-control-final-version.pdf
- As mentioned above, Google are increasingly limiting the collection of background Location data on Android. https://android-developers.googleblog.com/2020/02/safer-location-access.html
- 3. A recent example of a Premises owner (UK Network Rail) ‘collecting’ consumers WiFi Location Data. https://www-bbc-com.cdn.ampproject.org/c/s/www.bbc.com/news/amp/technology-51682280