Probabilistic links connect a device to a person algorithmically, based on characteristics and metadata such as:
Let's take a look at how probabilistic linking works. In this example, Person A browses to a news site on their tablet and then later from a desktop computer. While browsing, Person A does not log on to the news site. During each separate visit, the tablet and desktop share the same IP address.
Based on this information, the Device Graph evaluates IP address sharing patterns between both devices and links these devices together if the results suggest they belong to Person A. The end result is hierarchy of identity derived from algorithmic probability calculations.
In this example, the Device Graph linked both devices after they were used to access the same news site. But, devices do not have to be seen on the same site to be linked. To illustrate this point, let's say each device in this example visits completely different websites. The Device Graph algorithm can still make a probabilistic link based on their shared IP address and from an analysis of other data. This process is what helps make probabilistic linking so powerful for members of the Experience Cloud Device Co-op.