We have become a culture obsessed with making data-driven decisions. From sports, to business, to government, we strive to measure everything around us. In some cases accurate and detailed data is easily available, like clicks and engagement on a website. In other situations, like understanding people’s preferences, we go to great lengths, using proxies and fancy math to approximate a directional trend. Fortunately, the more things become digitized, the more data becomes available and the better decisions we can make.
Nowhere is this trend more noticeable than in the location space.
The average person in the United States visits 4.3 places a day (including home and work). That is a lot of movement on any given day, but up until very recently it was nearly impossible to quantify and understand.
As an example, in order to approximate foot traffic to fast food restaurants, Nielsen Scarborough had to collect over 200,000 surveys asking consumers where and when they visited a restaurant. Aside from the usual survey issues, like how many people actually remember the details of their fast food visits, surveys only capture a moment in time. Using this method it is nearly impossible to know what time the respondent left to get to the restaurant, how far he or she traveled to get there, how long they spent inside, where they went after, if they came back, and all sorts of other highly relevant data points.
The difficulty of accessing local data has been changing with the rise of smartphones. All of a sudden, it is possible to measure movement and context. We no longer have to guess and approximate where consumers go, because mobile phones can provide data that paints a much richer picture of where, when, and why users visit the world around them.
This new data source represents a massive opportunity for businesses that are trying to make more informed decisions. But there are critical caveats that buyers of this data need to understand.
This dataset is highly sensitive and significant efforts need to be taken to ensure privacy compliance. Whatever data set you use, it is paramount to look deep under the hood to ensure proper privacy handling. Specifically, this data should only come from users who understand and consent to sharing their data; it must be completely anonymized, the data must be stored securely; and it must only be shared at an aggregated level.
Source of Data
Making decisions based on location data requires a high degree of conviction that the insights are correct. Unlike in advertising where it is a game of numbers and the bar is to be right more than wrong, building intelligence and insights products requires significantly higher thresholds. As has been widely discussed here, here, and here location data from the exchanges is flat-out wrong more than it is right. Even when it is correct it is only precise to within 100 meters of the users’ actual locations. As you evaluate data sets, pay special attention to the source of the location data and make sure it is being pulled directly from the phone. For even better accuracy, look for partners that incorporate other pieces of data from the phone like accelerometer, gyroscope, wifi, and barometer.
Depth of Data
There is a widely held, but completely incorrect belief that a single lat/long pulled at the highest quality is enough to accurately determine location. There are several problems with isolated location pulls. The first is that it is impossible to tell if the person was actually visiting a location or just passing by. Multiple location pulls are needed to understand if the person is actually in McDonalds or just driving by, but that distinction significantly impacts the insights you can draw from the Lat/Long. The second issue is that GPS readings are flaky and you need more than one to triangulate more precisely where the person is.
Continuity of Data
A final, but critical point: in order to truly understand its customers companies cannot rely on sporadic data points. The true value in geo is understanding movement, not location. Ad exchanges, for example, can only determine users’ locations when they open an app that generates an ad request. This happens only intermittently, so relying on this data means one only sees random events. Continuous location tracking, on the other hand, allows for a much more granular understanding of consumer behaviour. It enables a much deeper exploration of things like frequency of visits and whether people with longer work commutes tend to have shorter shopping trips.
We live in a world where data is becoming more widely available and nowhere is this more true than with regard to location. Not all data is created equally though, and ensuring proper privacy compliance, and high-quality source, depth and continuity of the data is critical. With the right movement data you can make better decisions and transform your business by measuring and quantifying the world in a way that was never possible before.
Eli Portnoy is the CEO and co-founder of Sense360, which helps businesses make better decisions through sensor-data. He was previously CEO and co-founder of location-based ad network Thinknear, which he sold to Telenav in 2012. You can follow him on twitter @eportnoy.