The explosion of mobile data, and specifically mobile location data over the past few years, has brought about an incredible opportunity for businesses. While data has been used in making business decisions for centuries, the type of data and its volume is changing the way we do business. Within the marketing organization, for example, direct marketing has heavily relied on location data – or simply said, household address – to drive performance for decades. However, the use of data today goes well beyond the targeting use case – it’s being used more broadly to answer question across business units and across verticals. As a technologist, it’s important to better understand the dataset with which you’re working.
There are two important factors in understanding location data. One is the quality of the data, and the other is how to apply the data to gain actionable insights. Quantity of data is important, but without quality and thoughtful processing, having a large amount of raw data is like playing a bunch of musical notes in random order – it creates noise, rather than music. Just as it takes an expert composer to put together musical notes into masterpieces, it takes data scientists to apply machine learning to turn raw data into actionable business insights.
To make beautiful music, one needs to understand musical notes. To determine what, when and how to use location data to inform business decisions, it’s important to understand the different types of data available in the marketplace:
- IP-based: Guesses the location based on an IP Address – often very different from the true location of a user.
- Cell-sector triangulation: Uses cell tower data to triangulate where a device may be based on signal strength and other parameters. It’s generally acceptable in determining a very coarse location such as zip or city.
- Wi-Fi based Location: Uses Wi-Fi signals to infer location, typically within 10-100 meters; helpful in areas with higher signal density such as malls, buildings or urban environments.
- Global Positioning System (GPS): Utilizes Global Positioning Systems to pinpoint a latitude/longitude and can be both precise and accurate under the right conditions.
These data sets are often combined and converted to latitude/longitude for use in existing technology stacks. Because of this, data may be dirty due to the technology used to collect and convert it, or it could even be fraudulent. In fact, 90% of location data coming from publishers, often IP or cell-sector triangulation-based, can be inaccurate or fraudulent. This leads to a lot of noise.
This fact – that not all data is of equal value – led UberMedia to create extensive algorithms to screen what is often passed off as GPS or latitude/longitude data. The differences in not only the quality of the data coming in but also the ability to determine which data is real and which data is either fraudulent or duplicative can result in a huge disparity in performance. By applying various proprietary screening and filtering algorithms, the accuracy of the overall data set is exponentially improved. The bottom line is that if you take garbage data in, you get garbage data out. Ensuring you have a sense of the data cleansing techniques your partners use is critical to building a stronger, more accurate foundation to inform more impactful business decisions.
Today, location data is mostly used in the marketing and advertising space, but its application is spreading quickly into a wide variety of industries. For instance, retailers, quick service restaurants and shopping malls are leveraging location data to make site selection decisions, measure foot traffic trends, do competitive analysis, derive customer insights and more. Innovative cities are using data to do city planning, tourism studies and improve transportation. Other domains, like financial institutions, emergency response services, direct marketers, entertainment industry, tourist destinations and theme parks, are all starting to add mobile location data as an input to their business intelligence system. Using location data is like learning to master a new instrument. As businesses become more knowledgeable and skilled in this rich, expansive and valuable data set, there will be new opportunity to create use cases and masterpieces that we can’t even imagine today.
Contributed by: Gladys Kong, CEO of UberMedia. Since joining UberMedia in 2012 as CTO, Gladys has been responsible for assembling a best-in-class data science team and pivoting UberMedia from a social media app developer to a leading mobile advertising technology company. Gladys is an entrepreneur and founder of multiple tech companies, holding numerous patents in the mobile technology space. In 2016 and 2015, Gladys was named one of Business Insider’s “30 Most Powerful Women in Mobile Advertising.” In 2016, Mobile Marketer named her one of the 25 “Mobile Women to Watch”. Gladys holds a B.S. degree in Engineering and Applied Science from California Institute of Technology and an M.S degree in Computer Science from UCLA.
Sign up for the free insideBIGDATA newsletter.