Setup steps
Follow these steps to set up identity resolution from start to finish.
Step 1: Prepare your data to build an identity graph
- Go to Identity Resolution and click Add identity graph.
- Choose your data warehouse (Snowflake, Databricks, or BigQuery).
- Select the activation models you want to resolve across.
- Map columns from input models to identifier types which will be used for matching.
- Rank identifiers to deterministically match on.
Who does this? Platform admin, data or analytics engineer
Step 2: Set up Golden Record
- Go to the Golden Records tab.
- Add columns and select a survivorship rule for each (recency, frequency, etc.).
- Save and run the model to generate one trusted row per user.
Who does this? Platform admin, data or analytics engineer in consultation with marketer
Step 3: Choose your matching strategy
- Deterministic matching (Exact) is enabled by default.
- Toggle on Probabilistic Matching if you want fuzzy matching.
- Select the confidence levels you want to use (Strict, Loose).
Who does this? Platform admin, data or analytics engineering
Step 4: QA your matches
- Open the Summary tab to check match rates and identifier coverage.
- Use the Profiles tab to inspect merged identities.
- Spot-check specific users to confirm expected behavior.
Who does this? QA, marketer, and data or analytics engineer
Step 5: Use in syncs or audiences
- Build segments based on Golden Records
- Sync to tools like Braze, Salesforce, or ad platforms.
Who does this? Marketer
Setup checklist
Use this checklist to assign roles, follow key steps, and get Identity Resolution running for your team.
| Step | Article | What You’ll Do | Who Should Do It |
|---|---|---|---|
| 1 | Prepare your data → | Create a model and define relationships between source tables | Platform admin, data or analytics engineer |
| 2 | Golden Record Setup → | Define survivorship rules and select which values should be surfaced as canonical | Platform admin, data or analytics engineer in consultation with marketer |
| 3 | Choose Matching Strategy → Deterministic or Probabilistic | Enable deterministic or probabilistic matching and configure thresholds | Platform admin, data or analytics engineer |
| 4 | QA Your Matches → | Use the Summary and Profiles tabs to validate identities before activation | QA, marketer, and data or analytics engineer |
| 5 | Use in Syncs or Audiences → | Build audiences or sync profiles to downstream tools | Marketer |
Before You Launch
- Make sure your IDR model has run and populated
- Spot-check Profiles to confirm matches look accurate
- Double-check your Golden Record outputs
- Sync to a staging environment first, if available
- Confirm that destination tools (like Braze, SFMC, etc.) can accept new identifiers