The powerful ways in which technology enables individuals, businesses and organizations to use location information are more tantalizing and troublesome with every passing day. Typical location data is often deceptively simple: coordinates of latitude and longitude, or a common street address. Their relative simplicity contributes to their being easy targets of location fraud: swap around or add a few numbers and it could go unnoticed. More often, though, it’s not deliberate tomfoolery but just lousy luck and sloppiness with digital errors that propagate and persist, such as the Kansas site that can’t shake its alter-ego. Understanding how location data works: its sources, its formats, its limitations and its revelatory power, make all the difference for realizing its potential ROI.
You Want to Be Found, NOW
You’re in luck! There’s never been a better time to let the world know your whereabouts, 24-7. Here’s what you need to do: Keep your smart phone in your pocket at all times. Make sure that location services are always turned on, across the board for all apps. When you get somewhere, check in with the internet. When you’re moving between places, always navigate with online maps. Consider attaching mini-GPS trackers to your car and your computer, and wear one on your body. Don’t leave home without one!
You Don’t Want to be Found, NOW
Using our personal digital devices isn’t the only way to generate a digital footprint but it is one that readily gives us away. Don’t want to be found? Turn off your location services and manipulate your map. Fake out your browser. Mask or confuse the digital trail that your computer leaves by using a virtual private network or other methods of masking your IP address.
Companies Want to Know Where You Are, NOW
Meanwhile, the same technology that lets your find (or hide from) your friends or enemies is also being used by your credit card company to know where you are – and where you’re not – and thus monitor your account for fraudulent activity. For a bank to decline a purchase at the point-of-sale, several geographically-based factors come in to play concurrently, each involving some time and space signal that reflects an unexpected pattern. For example, the digital signal itself took longer than expected to arrive from the usual place of purchases, such as your home, if that were the typical site. The banking system must make those risk calculations in milliseconds so they rely on anticipated behavior and movement patterns. Knowing that we keep our phones close to ourselves, the location of the phone and the place from where we make a purchase should be the same. To scrutinize sale patterns after a transaction has been posted, fraud analysts also investigate purchases that were made where the phone was not.
Understanding the accuracy and uncertainty inherent within location information
Having location information that is as “correct” as it is wanted or needed to be is what makes this subject expensive, fraught with risk and surprisingly challenging. Eighty percent of all data may have a locational component (or not), but most of those locations are wrong. For example, Telenav’s ThinkNear produces a Location Score Index that serves as a high level, report-card style assessment of the location data used in mobile advertising. They regularly find that less than a third of the data sampled is situated accurately, regardless of whether the data are supposed to be within 100 m of a target or 100,000 m of a target. A business purchases what they think is a list of addresses of potential customers and only later realizes they have ZIP code centroids instead. These are not trivial matters when billions of advertising dollars are at stake each year, and costs for the risks to human safety and wellbeing are incalculable.
Whether the data inconsistencies are the result of sloppy or rushed data collection, misunderstandings about the limitations or functionality of the technologies, or actual deliberate fraud of one type or another, the temporal nature of the spatial information is one of the most important but most poorly understood contributing factors. To capture accurate information about people, or their cars, computers, or phones, that are in motion require data pulls at interval frequencies that are inconvenient, unrealistic or excessively expensive. The more dynamic the entities being tracked, the more opportunities for error and uncertainty, but this is the heart of mobile advertising: to be able to customize and target the content for just that person at just that time in just that place.
Static, stable entities bypass some of these challenges. Companies like AggData compile address locations of stores for hundreds of different retail chains and sell these as easy-to-use spreadsheets with geocoded point data ready to use. By going directly to the primary source for the information (think Starbuck’s own Find a Store), the risk that the location data itself is incorrect or not current is minimized. As AggData’s Chris Hathaway notes, retailers will naturally have their shared, public data be as complete, current and accurate as possible, in their own best interests.
So it’s an interesting time for this domain. A mix of well-intentioned data publishers and a few fraudsters exist, racing to develop and deploy technologies that leverage our demand for location-enabled goods and services while at the same time studying our exact purchasing and movement patterns. The location spoofing that allows a victim of domestic violence to keep herself safe is the same location spoofing that bogus marketers use to sell trumped-up addresses. I need my phone’s location known so I can find it, and I want to hide my phone’s location so no one else can find me. We’re located precariously on both sides of a double-edged sword.
Companies dealing with location data find themselves explaining basic geographical principles in their need to educate their clients. For example, both Placecast and Factual have illustrated how the decimal places of latitude and longitude coordinates translate to the geographic scale of what is being measured in distances. Want to set up some effective real time geofencing? Use anything fewer than four decimal places and you’ll probably miss your intended target. Well, if more decimal places is better, how about this guy who says his geocoded data will include no fewer than 8 decimal places! Brilliant! More is always better, so definitely go with his bid, and then remind me to tell you about this swampland I have in Florida for sale. Location, location, location!
Like I said earlier, location information seems deceptively simple. We first encounter latitude and longitude in elementary school but misconceptions about the system persist, and even GIS professionals will hesitate when asked whether the latitude value or the longitude value is plotted along the X or the Y axis of a graph. (Latitude are on the Y; Longitude are on the X. Yes, that’s correct.) Being a savvy user of location data means being prepared to solve problems, or at least recognize problems, and perhaps avoid having simple issues become a problem in the first place. The extent of the persistent accuracy problems with location data is strong evidence for persistent shortcomings in both geographic and quantitative literacy.