Identity Resolution for Beginners: How Data Cloud Knows Two Records Are the Same Person

Of everything Data Cloud does, one step feels like genuine magic to beginners: it takes a pile of records scattered across systems — different names, different spellings, different IDs — and works out which ones are the same human being. This is identity resolution, and understanding it is the difference between using Data Cloud and actually getting it. The good news: the magic is just careful, sensible logic, and I can walk you through it.

If you haven’t met Data Cloud yet, start with What Is Salesforce Data Cloud? — this post zooms into its most important step.

The problem, made concrete

Imagine three records from three systems:

• CRM: Robert Johnson, rob.johnson@gmail.com, phone 555-0142
• Web analytics: Bob Johnson, no email, phone 555-0142
• E-commerce: R. Johnson, rob.johnson@gmail.com, no phone

A human glances at these and thinks, “obviously the same guy.” But software doesn’t have intuition. The names don’t match (Robert vs Bob vs R.). It has to be told how to reason. Identity resolution is how you teach it that reasoning.

Step 1 — Match rules: “what makes two records the same?”

You define match rules — conditions that, when met, mean “these two records are probably the same person.” For our example:

• Rule A: same email address → match.
• Rule B: same phone number → match.

Apply them to our three records:

• CRM and E-commerce share the email → Rule A links them.
• CRM and Web share the phone → Rule B links them.
• Through CRM, all three are now connected — even though Web and E-commerce share nothing directly. They’re linked transitively, through the record in the middle.

That chain-linking is the quiet power of identity resolution: records with nothing in common can still be unified because a third record bridges them.

Step 2 — Matching isn’t always exact

Real data is messy, so match rules come in flavors:

• Exact match — the values must be identical (same email, character for character). Safe and strict.
• Fuzzy match — allows for human variation: Bob ≈ Robert, Jon ≈ John, minor typos, formatting differences in phone numbers.

Fuzzy matching is powerful but dangerous — too loose and you’ll merge two different people who happen to share a common name. The craft is tuning rules tight enough to be safe and loose enough to actually catch the variations. Beginners should start strict (exact matches on strong identifiers like email) and loosen carefully.

A strong identifier (email, phone, account number) is worth a hundred weak ones (first name, city). Build your match rules on strong identifiers and treat weak ones as supporting evidence, never as the sole reason to merge.

Step 3 — Reconciliation: building the one true profile

Once records are matched into a group, a new question appears: they disagree. One says the email is rob.johnson@gmail.com, another has none. One name is Robert, another Bob. The unified profile can only hold one value per field — so which wins?

That’s reconciliation. You set rules like “prefer the most recently updated value” or “trust the CRM source over the web source for names.” The result is a single unified profile — one clean, best-guess record representing the real person, assembled from the best pieces of every source.

Why this is the whole ballgame

Every valuable thing downstream depends on this step being right:

• A segment of “high-value customers” is only correct if each person is counted once, not three times as three fragments.
• An Agentforce agent grounded in customer data is only trustworthy if it sees the whole person, not a sliver.
• Personalization that treats Robert, Bob, and R. as three strangers will feel broken to the one human behind them.

Get identity resolution right and everything built on top inherits that quality. Get it wrong — merge the wrong people, or fail to merge the same person — and every downstream feature quietly inherits the error. This is why I tell beginners to respect this step: it’s small in the interface but enormous in consequence. As a musician I learned that one wrong note changes the whole piece; identity resolution is exactly that kind of detail.

Your next step

Now that you understand how a unified profile is built, learn how it’s used: Your First Segment in Data Cloud takes you from a clean profile to a usable audience. And keep the bigger picture in view with the Data Cloud category.