Header Background
Smartphone, laptop, smartwatch, and tablet connected by metallic threads converging at a glowing amber node on a slate desk.

How does a cross-device identity graph connect user identifiers?

Brian Dayman

A cross-device identity graph connects user identifiers by linking the many signals a person generates across their devices, browsers, and apps into a single, unified profile. Rather than treating each interaction as belonging to a separate anonymous user, the graph recognizes that the same individual is behind multiple touchpoints. The sections below unpack how that linking works in practice.

What types of identifiers does a cross-device identity graph link together?

A cross-device identity graph links both personal identifiers and device-level signals into one coherent record. Personal identifiers include email addresses, phone numbers, and physical mailing addresses. Device signals include cookie IDs, mobile advertising IDs, IP addresses, and browser fingerprints. Together, these data points form the connective tissue of an identity graph.

Identity graph data sources generally fall into two broad categories:

  • Authenticated identifiers: details a person voluntarily provides, such as an email address at login or a phone number during checkout
  • Anonymous signals: device-generated identifiers like mobile ad IDs or cookie values that exist without a name attached

The power of a persistent identity graph comes from bridging these two categories. When an authenticated identifier is observed alongside an anonymous device signal, the graph can associate that device with a known individual, extending recognition across sessions and channels without requiring the person to log in every time.

How does deterministic versus probabilistic matching work in an identity graph?

Deterministic matching links identifiers using exact, confirmed data points, while probabilistic matching infers connections using statistical likelihood. Both methods are used in identity graph marketing, and most real-world graphs rely on a combination of the two to maximize coverage without sacrificing accuracy.

Deterministic matching occurs when a person uses the same login credential across two different devices. The graph sees the same email address on a laptop and a smartphone and draws a direct, high-confidence link. No inference is needed because the match is exact.

Probabilistic matching handles the gaps. When two devices share similar behavioral patterns, the same household IP address, or overlapping browsing windows, the graph calculates the probability that both belong to the same person. The confidence level is lower than a deterministic match, but probabilistic linking dramatically expands the reach of a cross-device identity graph, especially in environments where authenticated signals are sparse.

What happens to a user profile when new device signals are detected?

When a real-time identity graph detects a new device signal, it evaluates whether that signal can be connected to an existing profile. If a match is found, the profile is updated immediately to include the new identifier. If no match exists, a new node is created within the graph and held until future signals allow it to be linked.

This process keeps the identity graph dynamic rather than static. As a person moves between devices throughout the day, each new signal is assessed against the existing graph in real time. The result is a profile that grows more complete with every interaction, without requiring manual data reconciliation after the fact.

A few important things happen during this update process:

  • Existing attributes on the profile, such as audience segments or enrichment data, are inherited by the newly linked identifier
  • Conflicting signals are weighted by confidence score rather than simply overwriting previous data
  • The updated profile becomes immediately available for downstream activation, such as personalization or frequency capping

How FullContact helps you build a real-time identity graph

We built our Resolve platform specifically to handle the complexity of building an identity graph at scale. Our identity graph connects authenticated and anonymous identifiers across devices using both deterministic and probabilistic matching, delivering resolved profiles through a real-time API in under 150 milliseconds. The platform appends 900+ personal and professional insights to each record, enriching profiles the moment a new signal is detected. We do all of this without taking ownership of your data, so your first-party signals stay yours. If you want to see how a persistent, real-time identity graph can unify your customer data, contact us and we will walk you through it.

Related Articles