Data 360 Tools · Step 2 of 2
Identity resolution, visualized.
Build the match rules and watch customer records resolve into Unified Individuals, the way Salesforce Data 360 (formerly Data Cloud) does it. Start with the demo, or load up to 50 records of your own. The records here are already in one shape; getting them there is data harmonization.
Load up to 50 records and watch your own data resolve. This is a visualization tool for learning and testing rule ideas, not a production matching engine. Your file is read in your browser, never uploaded or stored, and it disappears when you leave the page. Please use sample or anonymized data.
Known Person ID (optional). If you already know which rows belong to the same person, from an MDM golden record ID, a customer number, or your own hand labels, include it. Unlike a loyalty or party ID, it is never used for matching; the simulator uses it only as an answer key to grade your rules, never to match records. Without it, profiles that combine different names are flagged for review instead.
Records match if any rule matches, and a rule matches only when all of its criteria do. Click a criterion to cycle it.
How matching works here
- Exact matches regardless of case, and nothing else, like Data 360's Exact method. Data 360 also offers a Case Sensitive option for values like 15-digit CRM IDs.
- Normalized is Data 360's Exact Normalized method for email, phone, and address: it cleans up spacing, formatting, and special characters first.
[email protected]matches[email protected],555.555.0142matches(555) 555-0142, and 10 North Main Street Apt 4 matches 10 N Main St #4. - Fuzzy catches near matches in first or last name. In Data 360 it's an AI model that recognizes misspellings, accents, and synonyms, with High, Medium, and Low precision settings. Here it's a simple stand-in: nicknames (Bob and Robert, Jon and John or Jonathan), accents and punctuation (José and Jose, O'Brien and OBrien), one dropped or added letter, two swapped letters (Micheal and Michael), and sound-alikes (Luis and Luiz). It won't treat Mark and Mary as the same name.
- Loyalty ID is an exact match on a Party Identification number. It's strong evidence, but a card shared by a household merges everyone who uses it. Salesforce suggests adding the name to the rule if ID quality is in doubt.
- Blank values never match. Data 360 has a Match on Blank option; it's off here, as it is by default.
- Matching is transitive: if A matches B and B matches C, all three become one profile.
- The unified name follows a reconciliation rule: Most Frequent, or Source Priority (trust one system first). Reconciliation only decides what the unified profile shows; it never changes source records. Contact points aren't reconciled: every distinct email, phone, and address stays on the profile. Data 360 also offers Last Updated, which needs timestamps and isn't simulated here.
A simplified model for learning, not Data 360's actual matching engine.
Load a CSV above to see your records here.
Which rules are right?
It depends on what the profiles are for. The score above assumes one profile per person, which is the right target for service and billing, but not always for marketing.
Strict
Billing, healthcare, financial services
When a wrong merge shows one person's balance or diagnosis to someone else, a duplicate record is the cheaper mistake. Accept some duplicates to keep false merges near zero.
Balanced
Service, sales, and AI agents
A rep or an agent needs one real human per profile. Anchor every rule on the name, then let a contact point confirm it.
Relaxed
Marketing and household reach
A duplicate mailer costs cents, and treating a household as one unit is often the goal. A separate household ruleset handles that better than loosening the person-level rules.
Matching only works on what was harmonized first: split names, phones mapped as contact points, loyalty IDs that point at a person. See data harmonization, visualized →
Linking isn't merging: every record and contact point survives resolution. Why Data 360 doesn't build a golden record →
Profiles then get activated, and that's where counts shrink again. See why activations shrink →
That covers structured data. For the unstructured half, how policies and manuals become answers an agent can use, see vector search, visualized →
All demo data is fictional, and anything you load stays in your browser. An independent teaching tool, not affiliated with Salesforce; matching logic is simplified for learning. More Data 360 tools
Want the reasoning behind rules like these? Start with planning for identity resolution and the keychain analogy.