Deterministic identity resolution is a high-trust approach that relies on first-party data when it is known for certain that a user has taken an action. On the other hand, probabilistic identity resolution uses predictive algorithms to make educated guesses about users when information is not evident in deterministic data. This type of resolution can be used to add more value to deterministic data sets and to scale deterministic data models. Probabilistic identity resolution involves using predictive algorithms to determine the probability that two records belong to the same individual.
This process is based on an identity graph, which is a collection of methods used to associate information with a person, learn more about them, and reach them across devices and channels. It can also be used to fill in the gaps in deterministic data sets, providing more accurate information. Identity resolution software integrates consumer identifiers across all channels and devices in a precise, scalable, and privacy-compatible way to create a persistent and addressable individual profile. This software can be used to show the accuracy of certain fields in terms of their content, as well as which anonymous values (or equivalences) can easily match in different sources when analyzing duplicate records.
Many people mistakenly believe that CDPs (Customer Data Platforms) resolve identity. However, CDPs are not the only way to do this. With the phasing out of common identifiers like cookies, marketers have had to look for alternatives. Fingerprinting is one such alternative that is often done by advertisers or ad technology companies when they want to create persistent identifiers without the knowledge or approval of individuals or publishers.