This is the third article in our ongoing series “Under the Hood”, where we explore how the mobile location data ecosystem works, and what is required to turn that information into a precise understanding of physical world behavior.
The Maps that can Misguide Context
In our personal lives, we rely heavily on the maps and apps on our mobile devices to navigate the world with ease. Having this convenience at our fingertips makes life easier, and the maps feature on a mobile device can guide us through unfamiliar territories without ever missing a turn or having to pop a U-ie. We can all recognize that relying so heavily on one map’s view of the world can sometimes be misleading. Businesses close or change hands, retailers can be rebuilt or rezoned. Most often, we are not led astray by the location signals accuracy and find ourselves exactly where we want to be. But in those instances when you find yourself standing on a street corner, looking at your phone, spinning in circles because the blue dot says the restaurant is on this corner when it is nowhere to be seen, know your phone is not to blame; it’s the inaccuracy of the map you’re using. Driving directions or finding that restaurant is one thing, but what happens when revenue-impacting business decisions are based on the context of location according to only one map? Trusting only one map could result in leading your marketing budget down a dead end road.
Filtering Out the Faulty Data
At NinthDecimal, we know how important precise location data is. We never use IP Addresses as a primal understanding for a precise location because we know it can represent the center of a zip code, state, or even the country (like Potwin, Kansis, the center of the US, aka a centroid).
Step 1: Knowing that only 60% of inbound data is the type we want, we use several filters to eliminate cached IP Addresses, centroids, any signals less than 5 decimals. We only process data inside The Drift Zone.
Step 2: We study the length of user sessions to understand when in the session is prime-time for accurate signals. We do this to refine what is known as horizontal accuracy (how we know the “blue dot” is one meter vs. 100 meters). We’ve created a robust process to understand the Blue Dot Effect ingesting location services data makes it the most precise.
Step 3: Multi-geocoding process to determine the context of a location. We use several, ever-refreshing map databases to cross-reference the address, boundaries of a location, and the category of this entity (restaurant, office, park). Most companies only partner with one map database, or worse, use their own uncorroborated understanding of places.
Only after this process has happened does location data truly become physical world behavioral data and worthy of making decisions with.
Importance of Filtered Location Data
But what’s the worst that can happen when building context around unfiltered location data?
In December 2015, a tornado ripped through a Texas down, decimating parts of a neighborhood and partially destroying one woman’s home. She had made the choice to repair the damages rather than bulldoze the structure. Unfortunately for her, when the demo company plugged in the address of a different house intended to be leveled, the directions led to her home. The homeowner received a distressing call from her neighbor that her home had been totaled.
Which bring us to a case of a major IP map mishap. For over 10 years, a family living in Potwin, Kansas has been living with constant harassment. Their troubles range from being accused of identity theft to the FBI showing up to their home numerous times seeking criminals, police searching for children who were reported missing, and received a slew of threats from victims of online fraud…all because there are over 600 million IP addresses assigned to the latitude and longitude of where their home is. Potwin, Kansas happens to be in the dead enter of the United States.
This is why at NinthDecimal, we aren’t only looking at one map to understand the context of the location. We cross reference several of the largest independent and most trusted place databases for true physical world understanding. Most location data based companies are only placing devices in context of a location based on one places database. Even worse, sometimes that place database is their own. Without checks and balances against place databases, the ability to ensure accurate contextual understanding is drastically diminished.
For some companies, it is too much of an obstacle to validate and contextualize inbound data because their technology and geo-coding processes haven’t been architected to parse out the imprecise signals and validate that you actually are where your device thinks you are. For NinthDecimal, it’s necessary to transform location data into physical world understanding; because making revenue-impacting marketing decisions deserves a high fidelity foundation.