Unified People profiles are produced by an identity resolution ruleset in Data Cloud. You do not create a People record by hand. You define the rules, then Data Cloud builds and maintains the unified profiles for you. Here is the shape of that configuration.
- Map data to the Individual DMO
Bring your CRM, marketing, and commerce sources in as data streams, then map their person fields to the Individual data model object and the related Contact Point objects. Clean mapping here is what makes later matching possible.
- Create an identity resolution ruleset
In Data Cloud setup, create a ruleset targeting the Individual DMO. The ruleset is the container that holds all of your match and reconciliation logic for People profiles.
- Add match rules
Define how two records count as the same individual, for example an exact match on normalized email or on phone number. Add normalization steps so casing and formatting differences do not break matches.
- Set reconciliation rules
For each attribute, choose how conflicts resolve: most recent value, most frequent value, or a preferred source. Tune address and contact fields separately from name and demographic fields.
- Run, review, and refine
Process the ruleset, then inspect the unified individuals it produces. Watch for over-merging and under-merging, adjust the rules, and reprocess until the profiles look right.
The identifiers that link records, such as email, phone, or a loyalty key. Looser criteria merge more aggressively; stricter criteria keep records separate.
Per-field tie-breaker for choosing a winning value: recency, frequency, or source priority.
Formula and standardization applied before comparison so values like email casing and phone formatting line up.
Whether multi-valued channels (email, phone, address) are kept as separate Contact Point records rather than collapsed to one.
- Over-merging is worse than under-merging. Two real people fused into one profile leaks personalization and consent across them, which is hard to unwind.
- Identifiers must be normalized before matching. Without it, JANE@EXAMPLE.COM and jane@example.com look like two people and profiles fragment.
- A unified individual ID can change as rules and data evolve. Do not treat it as a permanent external key for integrations.
- Reconciliation is per field, not global. A single recency rule for the whole profile usually produces strange results on at least one attribute.