Connect source data through shared business definitions

Connect records across systems while preserving their original definitions and source history.

Customer roles across systems

CRM account

A commercial relationship

The account a team manages may include several locations or legal entities.

Legal counterparty

A contracting entity

The party bound by an agreement has its own identity and effective relationships.

Billing recipient

An invoice recipient

The invoice recipient may differ from the contracting party, service user or delivery location.

The domain owner confirms which customer roles apply to the decision.
Define the business concepts

Define the business concepts

A semantic model expresses what the business needs to know about its entities and relationships. For a delivery exception, that may include an order, an item, a site, a commitment and an accountable team. Begin with those concepts rather than copying every table and column into a second schema.

Define each concept with the domain owner and retain the source context. Two systems may both contain a customer identifier while referring to different commercial roles. A useful model connects those roles explicitly instead of flattening them into an apparently universal record.

A data architect examines how a field with its source identifier maps to an agreed business definition.

Match records across systems

Choose identifiers that remain stable within the agreed scope. Preserve original source keys and distinguish an exact reference from an inferred match. If matching uses names, addresses or other attributes, document the evidence and the process for reviewing uncertainty.

Identity changes need a history. Mergers, renamed organizations, replaced assets and split material lots can alter relationships without erasing the past. Effective dates and explicit change records help users understand what was true when an earlier event or decision occurred.

Source identifiers

Preserve the source key, the system it belongs to and the scope in which it identifies the entity.

Proposed identity matches

Retain the matching evidence, uncertainty and the person who can confirm or correct the relationship.

Identity review compares customer keys with their source identifiers and retains a proposed relationship for steward review.
Source identifiers and matching evidence remain visible while the steward reviews the proposed relationship.

Define the relationships

A relationship needs meaning, direction and scope. An account owns a contract, a site receives an order and a component belongs to an asset during a defined period. State which source establishes each relationship and how conflicting evidence is resolved.

Define the relationships
Model elementDecision to record
EntityWhat real business concept does this represent?
IdentifierWhich authority assigns it and where is it unique?
RelationshipWhat does the connection mean and when is it valid?
MeasureWhat units, calculation and reporting scope apply?

Keep definitions current

A shared definition becomes useful through adoption, review and maintenance. Assign domain owners who can approve meaning and technical owners who can assess downstream effects. Review proposed changes against representative records and the workflows that consume the model.

Version meaningful changes and preserve compatibility where necessary. A revised definition of available inventory can affect decisions even when a field name stays the same. Communicate that change as a business rule update rather than treating it as an invisible implementation detail.

Test the model against real cases

Ask whether a user can explain a real case with the modeled relationships. Test ambiguous identities, missing links, historical changes and conflicting measures. A diagram can look complete while failing to represent a routine operational exception.

Check that users can trace each relationship to its source, identify unresolved matches and add a new source without changing established business definitions.

Talk to us about your operations.

  • Agree on business concepts and their source authority
  • Preserve identity, relationship and definition history
  • Check that the model answers real business questions
Discuss semantic modeling