Planning for Identity Resolution in Salesforce Data Cloud
Identity Resolution in Salesforce Data Cloud is powerful — but the work that happens before you configure a single matching rule is what determines whether it actually works. Two steps, done in order, will save you real time and platform credits.
Step 1: Profile the Source Data
Understanding your data quality is foundational. Before you build rules, answer some basic questions: Are essential fields like Name, Address, Email, and Phone actually populated? Do quality controls exist at your entry points?
The technique I recommend is running frequency distributions on critical fields — especially phone and email. This surfaces a category of problems that’s easy to miss until it’s too late:
- Duplicate placeholder values — dummy phone numbers or test email addresses that appear across hundreds of records
- Inconsistent formatting across sources
- High rates of missing data in the fields you’re planning to match on
Think of this as a data reality check before you build rules. You’re not trying to fix everything — you’re trying to know what you’re working with.
Step 2: Run a Dry Test in a Spreadsheet
This sounds low-tech because it is. Export a sample of your data and manually apply your matching logic before touching the platform.
The sample should include three categories:
- Records you expect to match
- Records you expect not to match
- Grey-area records you’re genuinely uncertain about
Working through these cases by hand does a few things that matter. It surfaces edge cases before you’ve spent platform credits. It establishes a baseline you can validate against once you run the real thing. And perhaps most usefully, it gives you a stakeholder-friendly explanation of exactly how Identity Resolution works in your specific data context — something that’s much harder to explain after the fact.
From There
Start with conservative matching rules. Compare your first run against the expectations you set in the spreadsheet. Adjust incrementally.
Identity Resolution done right isn’t a one-time configuration — it’s an iterative process grounded in understanding your data before the platform ever sees it.