Skip to content
FirstHive Logo FirstHive

13. Unification and Identity Resolution

FirstHive enables a robust identity resolution framework through a three- or four-tiered identity structure that unifies customer data across channels and devices.

Key Advantages

  • 360-Degree View: Builds unified customer identities across all marketing channels and interaction sources.
  • Real-Time Updates: Automatically updates profiles as new data is ingested.
  • Historical Tracking: Maintains a complete timeline of customer journeys and interactions.
  • Predictive Attributes: Uses AI to generate insights on customer preferences, intent, and behavior patterns.

Identity Structure

FirstHive creates a layered identity system to help track and unify user data across different channels and devices. Here’s how it works:

  1. Event ID: This is the base layer that tracks every single event (like clicks, form fills, purchases). It’s persistent and forms the foundation of identity. Web or mobile event Ids are also tagged with a corresponding session Id.
  2. Channel ID: This represents how the customer is contacted across various channels —email address, phone number, device ID, etc. It ties interaction data to specific communication channels.
  3. FHID (FirstHive ID): This is the main identity layer—a contextual, cross-channel, cross-device ID that unifies information across channels and devices to represent a single customer. It serves as the primary abstraction for targeting and personalized communication.
  4. Entity Identity ID: This adds another layer for brands needing complex identity management based on roles (e.g., agent vs. customer) or across multiple brands.

Supported Unification Models

Depending on your FirstHive license, you can choose from five models for merging customer data into unified identities:

  1. Basic Deterministic (FH Lite): Unifies identities only when exact matches are found on primary fields (e.g., same email or phone number).
  2. Deterministic Hierarchical: Similar to basic deterministic but uses priority structure across identifiers to resolve and merge profiles (e.g., email > phone > cookie ID) to resolve conflicts.
  3. Hybrid ML-Based (Default): Uses a combination of deterministic logic with machine learning to improve matching accuracy to merge profiles. This is the standard approach in most FirstHive plans.
  4. Hybrid ML-Based (Configurable): Same as above, but with the ability to customize the rules ML thresholds, rules, and data handling logic.
  5. Custom: A fully tailored model built for specific business needs or complex identity resolution requirements.

AI-Powered Identity Resolution

FirstHive is the first CDP globally to leverage machine learning for real-time, intelligent customer identity resolution, delivering higher accuracy and match confidence. This includes the Probabilistic and Biometric Identity Matching models available through the AI/ML Capabilities Model Marketplace.

Configuration Steps

Setting up AI-driven identity resolution in FirstHive involves three core stages. Each step ensures that your system can accurately identify and unify customer data across all integrated sources.

Stage 1: Identity Key Setup

Begin by defining the identifiers that FirstHive will use to recognize and match customer profiles. Which identifier ranks as “primary” is configured per account and per vertical — it is not fixed to any one field. Email, phone, and customer ID are common primary identifiers, but plenty of implementations lead with something else: a government or tax ID where one is mandatory for compliance, a loyalty or membership number where the source system pre-validates it upstream, or a mobile number where that’s the ecosystem’s universal anchor. Work with your Solution Engineer to define the right priority order for your data.

  • Primary Identifiers: Typically email address, phone number, loyalty ID, or customer ID — but confirm the right choice for your vertical and data sources
  • Secondary Identifiers: Device ID, social handle, IP, Tokens, etc.
  • Behavioral Identifiers: Purchase history, browsing behavior

For worked examples of how identifier priority differs by industry — including cases where the standard email/phone-first pattern doesn’t apply — see Events and Entity Schemas.

Stage 2: Configure ML-Based Matching Rules

Once identifiers are in place, configure how the system should interpret them using AI logic.

  • Configure AI confidence thresholds to determine when records should be auto-matched.
  • Enable fuzzy matching to catch near-identical but non-exact records (e.g., slight typos).
  • Define custom rules tailored to your business’s logic for more precise control over unification.

Stage 3: Decide Profile Merge Strategies

After matches are identified, decide how profile data should be merged into a single source of truth.

  • Data Prioritization: Set which system’s data takes precedence in case of overlap
  • Temporal Logic: Use the most recent value for fields like address or phone number
  • Manual Review: Route ambiguous or low-confidence matches to a review workflow