Identity Resolution (IDR) overview

Unify your data into clean, customer-level profiles that power every team’s workflow.

What is Identity Resolution?

Identity resolution (IDR) connects data from multiple sources—like web events, purchases, CRMs, and support tools—into a single view of a person. Instead of juggling multiple emails, phone numbers, or user IDs, you create a persistent identity for each person (a Golden Record).

With unified identities, you can:

Build accurate audience segments
Personalize messages across channels
Run analytics and modeling with clean inputs
Sync reliable records to CRMs and ad platforms

How identity resolution works

You can use two approaches to match records:

Matching strategy	Best for	How it works	Use cases	Data sources
Deterministic	Clean, structured data; typically collected through automated tracking	Links records using exact matches (like email == email)	Anonymous to known resolution, operations, transactional emails	App events, CRM, CDP
Probabilistic	Messy or inconsistent data; typically collected through human data entry	Uses machine learning, fuzzy logic, and similarity scoring to link similar records	General lifecycle and CRM marketing, paid media ads, analytics	Form fills, offline purchases, third-party data

You can use either strategy—or both—depending on your data quality and goals.

What you get

When your identity project runs, it creates:

Output	Deterministic	Probabilistic
A resolved identity graph (mapping identifiers and input records to a stable unified ID)	✅	✅
A Golden Record table (one row per person, with trusted values)	✅	✅
Confidence tiers (Exact / Strict / Loose) to filter by match quality	❌	✅

You can access these in your warehouse and in Customer Studio (if enabled).

Who uses identity resolution?

Role	What you do	What IDR gives you
Lifecycle marketer	Personalize and target users	Accurate and complete segments incorporating the full customer journey
Analytics lead	Run LTV, CAC, retention models	One row per identity with consolidated transactions across different duplicate customer records
Data engineer	Join noisy datasets cleanly	Configurable match logic
RevOps	Sync clean data to tools	Canonical traits and deduplication

Choose your matching strategy

Start with Deterministic Matching if you have clean, shared IDs across systems (email, user ID, device ID).
Add Probabilistic Matching when your data has typos, formatting issues, or ambiguous IDs (e.g. name + zip).
You can configure both within the same model, and filter downstream usage by confidence level.

Create a Golden Record

Your Golden Record is a single, flattened row for each user. It includes the most trustworthy value for each trait, selected using rules you define.

Why it matters

You avoid duplicates across systems
You sync clean traits to downstream tools
You get one row per user for reporting

Define survivorship rules

Rule	What it does
Recency	Uses the most recently seen value
Frequency	Uses the most common value
Source priority	Uses value from a preferred table
Array	Stores all unique values in a list

Example use cases

Sync the most recent email to Braze
Run LTV analysis with merged purchase history
Enrich CRM with trusted name, phone, and email

Review matches and data quality

Before you sync anything, QA your results:

Open the Summary tab to check match rate and identifier coverage
Use the Profiles tab to inspect merge logic and trait accuracy
Validate fields in the Golden Record
Confirm that thresholds are appropriate for your use case

What to look for

Task	What to check
Identifier gaps	Are important fields like name or email missing?
Unexpected merges	Check for incorrect joins, adjust thresholds if needed
Low match rate	Consider enabling probabilistic matching