Data 360 Series | Chapter 4 — From scattered fragments to one unified view

This is where the magic happens — and where most of the real work lives. Harmonizing and unifying data is the craft at the heart of Data 360. And it’s important to understand that a Unified Profile is not a golden record — it’s something subtler and more powerful than that.

If ingestion is about getting data in, harmonization is about making that data speak the same language, Unification is about recognizing that four different records across four systems might all be the same person.

Harmonization: mapping to the C360 Data Model

The Customer 360 Data Model (C360) is Data 360’s canonical schema — a pre-loaded, industry-agnostic data model with over 300 standard objects. When you harmonize data, you’re mapping your source fields to this shared language.

This mapping — DLO field to DMO field — is where the team spend most of their time. Get it right upfront, and everything downstream flows naturally. Get it wrong, and you’ll be retrofitting identity resolution rules and calculated insights for months.

Identity resolution: recognizing the same person across systems

This is the crown jewel of Data 360. Identity resolution does not pick winning values or overwrite source data. It creates a set of keys that unlock your source data — identifying all matching records that relate to the same entity.

Your source records remain intact, linked together via the Unified Profile’s UUID. You then choose which source system’s data to use for each specific business use case.

The pipeline runs through four stages:

• Candidate selection — Before comparing records, the engine uses Blocking Keys and Locality Sensitive Hashing (LSH) to narrow the field. LSH is a mathematical technique that groups records likely to be similar into “buckets” — so instead of comparing every record against every other record (which would take forever at billions of rows), it only compares records in the same bucket. Think of it like sorting mail by postcode before sorting by name.
• Deep matching — AI models calculate a probabilistic match score for each candidate pair, intelligently handling misspellings, variations, and formatting differences.
• Clustering — Matched records are grouped into clusters, including transitive matches (if A matches B and B matches C, all three link to the same profile — even if A and C were never directly compared).
• Reconciliation — Field values from clustered records are evaluated using defined rules (Most Recent, Most Frequent, Source Priority) to populate a representative excerpt on the Unified Profile. No source data is deleted or overridden.

Key Rings: the identity mechanism

Behind unified profiles is the Key Ring mechanism. Every Unified Individual has a UUID (a universally unique identifier). Attached to that UUID is a set of “keys” — one per source system identifier. Think of it literally as a keyring: each key can be added, removed, or transferred as your data evolves. All original records remain intact, linked to the UUID through identity link objects that map every source record ID to its corresponding Unified Profile ID.

Near-real-time processing and entity types

Identity resolution doesn’t just run once — it operates continuously. Small batches of changes can be processed as frequently as every 15 minutes, ensuring that the Unified Profile always reflects the freshest available data for personalization and Agentforce interactions.

The engine supports three entity resolution types, each serving different use cases —

Individual matching: linking personal identifiers across systems into a Unified Individual.

Account matching: linking company records for B2B scenarios, including fuzzy company name matching.

Household matching: grouping related Unified Individuals into households.

The golden record misconception:

Data 360’s Unified Profile is not a golden record. It does not propagate reconciled data back to source systems, merge records, or override existing values. It is a set of identity links — a key ring — that lets you choose which source system’s data to use for each use case. If you’re evaluating Data 360 as an MDM replacement, this is a critical design distinction.

Best practice: Only import active profiles into Data 360. Every record runs through identity resolution — which has the highest credit multiplier of any operation. Importing 5 million inactive customers just to exclude them later is a costly mistake.