AEP schemas: designing XDM schemas step by step

Key concepts

Class

Defines the behaviour of the data: record (profile attributes) or time-series (events).

Field group

A reusable set of fields added to a schema, either Adobe-provided or custom.

Data type

A reusable object structure (for example an address) used inside field groups.

Identity field

A field tagged with an identity namespace so Identity Service can link records.

Union schema

The combined view of all Profile-enabled schemas of the same class.

Step-by-step: design a schema

1. Start from the business use case and list the attributes and events it needs.
2. Pick the class: XDM Individual Profile for CRM-style attributes, XDM ExperienceEvent for web, app, transaction and interaction events.
3. Reuse standard field groups first; create custom field groups under your tenant namespace only for what is missing.
4. Identify identity fields (email, CRM ID, ECID, phone) and set one primary identity per schema where required.
5. Review naming, types and required fields with data owners before creating datasets.
6. Enable for Profile only when you are sure — Profile-enabled schemas can only have additive changes afterwards.

Example: retail purchase event

A purchase event schema typically uses the XDM ExperienceEvent class with the Commerce Details field group, an Environment Details field group for web context, and identity fields for ECID and a hashed customer ID. Order totals, product SKUs and quantities sit in standard commerce fields so that Customer Journey Analytics and Journey Optimizer can use them without custom mapping.

Best practices checklist

• Design for use cases, not for every column in the source system.
• Keep custom fields under one tenant object with consistent naming.
• Avoid storing personal data you do not need; apply data usage labels.
• Separate profile attributes from events — never put frequently changing events into profile records.
• Version and document schemas; treat breaking changes as a new schema.
• Validate sample payloads against the schema before production ingestion.

Who this guide is for

This guide is written for beginners, students and career switchers who want a clear mental model, and for developers, data engineers, implementation specialists, consultants, architects, marketers, agencies and enterprise teams who need practical, step-by-step detail they can apply on real projects.