Modern consumers interact with brands across an average of five or more devices and numerous digital channels. They might research a product on their phone during a commute, compare options on a work laptop, and ultimately purchase on a tablet at home. Identity resolution is the technology and process that connects these fragmented touchpoints into a unified view of a single person, enabling advertisers to deliver relevant messaging and measure campaign effectiveness across the full customer journey.
What Is Identity Resolution?
Identity resolution is the process of matching multiple identifiers and data signals to a single individual or household. In advertising, this means connecting cookies, device IDs, email addresses, phone numbers, and other signals to create a persistent identifier that represents one person across their various digital interactions.
Without identity resolution, each device or browser appears as a separate user. A person who visits a website on their phone and later on their laptop would look like two different people, leading to duplicated frequency, wasted reach, inaccurate measurement, and poor user experience. Identity resolution solves this by recognizing that these separate signals belong to the same individual.
Deterministic vs Probabilistic Identity Matching
Identity resolution technologies generally fall into two categories based on the confidence level of their matching methodology.
Deterministic Matching
Deterministic identity resolution relies on known, definitive identifiers to connect users across devices. The most common deterministic signal is an authenticated login. When a user logs into the same service, whether it is Gmail, Facebook, Amazon, or a publisher's website, on multiple devices, the platform can definitively link those devices to one person.
Deterministic matching offers high accuracy but limited scale. Only a portion of users authenticate on any given platform, meaning deterministic matching alone cannot resolve the identity of the entire addressable audience. However, the matches it does make are highly reliable.
Other deterministic signals include email addresses, phone numbers, and CRM data that users voluntarily provide. These signals are typically hashed (converted to anonymized strings) before being used for matching, providing a layer of privacy protection.
Probabilistic Matching
Probabilistic identity resolution uses statistical modeling to infer connections between devices and identifiers without definitive proof. The approach analyzes signals like IP addresses, device types, operating systems, browser configurations, location data, and browsing patterns to calculate the probability that two devices belong to the same person.
For example, if a smartphone and a laptop consistently connect from the same IP address, use the same Wi-Fi network at similar times, and visit overlapping sets of websites, a probabilistic model might assign a high confidence score that they belong to the same household or individual.
Probabilistic matching offers much greater scale than deterministic methods because it does not require user authentication. However, its accuracy is lower, with match rates varying significantly depending on the signals available and the sophistication of the model. False positives, where two devices belonging to different people are incorrectly linked, are an inherent risk.
Identity Graphs
An identity graph is a database that stores the connections between identifiers and creates a unified view of individuals or households. Identity graphs serve as the foundation for cross-device advertising, personalization, and measurement.
Identity graphs typically contain multiple layers of identifiers:
- First-party identifiers: Data from direct customer relationships, including CRM records, email addresses, loyalty program IDs, and authenticated session data.
- Device identifiers: Mobile advertising IDs (IDFA for Apple, GAID for Android), cookie IDs, and browser fingerprint signals.
- Offline identifiers: Physical addresses, phone numbers, and other data points that connect digital identities to real-world individuals.
- Probabilistic clusters: Statistically inferred device groupings based on behavioral and environmental signals.
Major identity graph providers include LiveRamp, which operates one of the largest deterministic identity solutions through its IdentityLink platform; Experian, which combines marketing data with its consumer credit database; and the major walled gardens like Google and Meta, which maintain their own massive authenticated identity graphs.
Identity Resolution Use Cases in Advertising
Identity resolution enables several critical advertising capabilities:
Frequency Management
Without cross-device identity, frequency caps only work within a single browser or device. A user might see the same ad three times on their phone, three times on their laptop, and three times on their tablet, for a total frequency of nine despite a cap of three. Identity resolution lets advertisers manage frequency across all of a user's devices, improving user experience and reducing wasted impressions.
Sequential Messaging
Identity resolution enables storytelling across devices and channels. An advertiser can show an awareness-focused video on a user's connected TV, follow up with a product detail ad on their phone, and serve a promotional offer on their laptop. Each message builds on the previous one because the advertiser recognizes the same person across touchpoints.
Cross-Device Attribution
When a user sees an ad on their phone but converts on their laptop, cross-device identity resolution ensures that the mobile ad receives appropriate credit. Without it, the mobile campaign would appear to have zero conversions while the laptop browsing session would be credited as an organic visit.
Audience Suppression
Advertisers can suppress ads to existing customers across all their devices, preventing wasted spend on users who have already converted. This requires identity resolution to connect the CRM record of a customer to their various device identifiers.
Privacy Challenges and the Future of Identity
The identity resolution landscape is undergoing fundamental change driven by privacy regulations and platform decisions. Apple's App Tracking Transparency framework has significantly reduced access to IDFA, the primary mobile identifier on iOS devices. Google's Privacy Sandbox initiatives are replacing third-party cookies with privacy-preserving alternatives. GDPR and CCPA require explicit consent for the data collection that many identity resolution methods depend on.
These changes are pushing the industry toward several emerging approaches:
- First-party data activation: Brands are investing in strategies to collect authenticated first-party data through loyalty programs, email registration, and value exchanges. This data is owned by the brand and not subject to third-party deprecation.
- Clean rooms: Data clean rooms like those from LiveRamp, Snowflake, and AWS allow advertisers and publishers to match their first-party data without either party exposing raw data to the other, enabling identity resolution within a privacy-safe environment.
- Privacy-preserving identifiers: Solutions like Unified ID 2.0 use encrypted, hashed email addresses as an open-source identity standard, requiring user consent and providing transparency about data use.
- Cohort-based approaches: Google's Topics API and similar technologies move away from individual identity entirely, instead classifying users into interest-based groups that can be targeted without individual tracking.
Building an Identity Strategy
Organizations developing an identity resolution strategy should start with their first-party data assets. Understanding what customer data you already have and establishing processes to collect more through consented value exchanges creates a foundation that is durable regardless of how third-party identity evolves.
From there, evaluate identity partners based on their match rates, accuracy, privacy compliance, and interoperability with your existing technology stack. The most effective identity strategies layer multiple approaches, using deterministic matching where possible, supplementing with probabilistic signals where appropriate, and maintaining flexibility to adopt emerging solutions as the landscape continues to evolve.
