How do we determine if a person's phone and home computer will be tracked as one or more persons?
In the world of technology, we understand things can be confusing. This question regarding a person's multiple devices is not exactly a 101 course. Nonetheless, we will try our best to explain.
First, let's define what an IP address is. An IP address is comparable to your postal code. Wikipedia elaborates this definition as a label that is used to identify one or more devices on a computer network, such as the internet. Essentially, it's via an IP address that the big vast internet knows how to send specific information to a specific computer or device.
Now, here's where an IP address differs from other forms of mobile unique identifiers: 1) IP addresses are not permanent. They are often subject to change when a user changes service providers. 2) An IP address can be shared across multiple devices. It's not uncommon to have multiple phones or devices and computers connected to one internet router with one IP.
A Device ID is a string of letters and numbers that is there to identify every device (like a smartphone or tablet) in the world. It is stored on a mobile device and is retrievable by any app that is downloaded and installed. Apps typically retrieve the hashed device ID for identification when talking to servers, but it is also helpful in identifying the movement of people's devices (such as phones) in your geolocation data.
Still, tracking with us? 😉
Because unique identifiers are not always readily available, Near (previously UberMedia) has developed a systematic procedure for associating device IDs with users called Device-IP bridging.
Here is an excerpt from their white paper:
Given the one-to-many relationship between IP addresses and smartphones, any connection to the two is probabilistic. This means the match isn’t an exact match, but is a best estimate of a match. Near performs this probabilistic matching by looking over a lookback period of device usage to determine the best device-IP matches when presented with either a set of IP addresses (from say, a pixel fire from ads placed on desktop web) to pair to devices or the reverse (a set of devices who have all been to a Walmart for instance).
Given the match is not deterministic, it is critical to the success of any bridging effort to account for the probabilistic match in the end product the bridged data fills in to. Near accomplishes this in the following ways:
• For media attribution, a carefully constructed partial attribution methodology is employed where the contribution to the final influenced visitor numbers and lift calculations account for the probability of ad exposure through bridging.
• For audience building, a careful consideration of the history of the device and matches to likely home IP addresses where the probability of a direct match are highest.
Need more help? Talk to a real human.