Identity Resolution

Identity resolution links a user's many identifiers into a unified profile. This module covers deterministic vs probabilistic, the major identifiers, and the quality measurement that prevents bad personalization.

1. Why identity resolution

A user appears in dozens of systems with different identifiers: web cookies, mobile IDs, email, phone, customer-ID. Identity resolution links these into a unified persistent profile. Without it, every marketing system has a partial view of the customer.

2. Deterministic vs probabilistic

3. Identity graphs

An identity graph stores relationships between identifiers. Each node is an identifier; each edge is a confidence-scored relationship. The graph grows as new signals connect previously-disjoint identifiers.

4. Common identifiers

• Email (highly reliable, persistent).
• Phone (highly reliable, persistent).
• Customer-ID (most reliable for known customers).
• Device IDs (mobile MAID; reliability declining).
• First-party cookies (browser-specific, expire).
• Third-party cookies (deprecated).
• Login session.
• IP address (low-quality identifier, especially with mobile).

5. Cross-device

Linking a user's desktop, mobile web, and mobile app. Methods: login (most reliable), email collection, deterministic match via shared identifier, probabilistic via behavior/location.

6. Cross-domain

Linking the same user across multiple owned domains. Methods: first-party cookie + login, server-side identity stitching, third-party graph (LiveRamp).

7. Online-to-offline

Linking online identifiers to offline customer records. Methods: email/phone capture, in-store login, transaction matching, third-party graph providers.

8. Privacy

• Identity work concentrates PII; treat with appropriate security.
• Consent requirements vary by jurisdiction.
• Right-to-deletion must propagate across the identity graph.
• Hashing PII for vendor sharing.

9. Vendor solutions

• LiveRamp: The leading commercial identity graph.
• Tapad (Experian): Probabilistic device graph.
• Adobe Real-Time CDP: Adobe-native graph.
• Salesforce Data Cloud: Salesforce-native graph.
• Acxiom (IPG Mediabrands): Marketing identity graph.
• The Trade Desk Unified ID 2.0: Email-based open identity.

10. Measuring quality

• Match rate: percentage of records that resolve to a known profile.
• Match accuracy: confirmed correct matches vs total.
• Profile completeness: data points per profile.
• Recency: how current the linked identifiers are.

11. Common mistakes

Part of the Customer Data Platforms series · RGM Training