Open Retail Data Model (ORDM)

The Open Retail Data Model (ORDM) is an open, Apache-2.0-licensed foundation of reusable retail data assets — canonical entities, semantic models, business glossary, and synthetic data — built together with retail partners, Databricks, and the broader retail community.

It gives retailers, partners, and solution builders a shared, ready-to-use starting point so analytics and AI solutions compose instead of being rebuilt from scratch for every project. Adopt it as-is, extend it, or adapt it to your own data landscape.

ORDM is the retail instance of the Industry Data Model — Open Program (DIDM-OP).

What's inside

Layer	Folder	What it is
Canonical core	canonical-core/	Conformed business entities — Customer, Product, Order, Inventory, Store, Transaction. The shared primitives every outcome builds on.
Outcome packages	outcome-packages/	One package per retail business outcome (12 in the first wave). Each bundles metric views, semantic metadata, and analytical views on top of the canonical core.
Synthetic data	synthetic-data/	Generators (Python + dbldatagen) and sample data so you can deploy and explore without your own data.
Examples	examples/	Deployment templates (Databricks Asset Bundle).

Architecture at a glance

                         ┌──────────────────────────────────┐
                         │  Canonical core                   │  conformed entities
                         │  Customer · Product · Order ·     │  (shared primitives)
                         │  Inventory · Store · Transaction  │
                         └──────────────────────────────────┘
                                          │
                                          ▼
                         ┌──────────────────────────────────┐
                         │  Outcome packages                 │  metric views + semantics
                         │  12 retail outcomes               │  (consumption-ready)
                         └──────────────────────────────────┘
                                          │
                                          ▼
                              Your data & extensions

You bring your own source data and ingestion; ORDM provides the conformed model and the outcome-aligned semantics on top. See docs/architecture.md.

The 12 retail outcomes (first wave)

Respond & Predict to Customer Behavior

• Unified Customer View
• Actionable Customer Understanding
• Agentic Commerce
• Commerce Media Networks

Profitable Volume Growth

• Balancing Margin and Volume
• Promote with Purpose

Collaborative Supply Chain

• Early Risk Detection
• Smarter Demand and Inventory Decisions
• Data Sharing with Suppliers

Intelligent In-Store Experience

• Connected Store Signals
• Predictive Operations
• AI-Assisted Store Associates

Getting started

1. Explore the model — browse canonical-core/ for the core entities and outcome-packages/ for outcome-aligned metrics.
2. Try it with synthetic data — use synthetic-data/ to populate the model in your own Databricks workspace, no source data required.
3. Deploy — the examples/deploy-with-dab/ Databricks Asset Bundle deploys the canonical core and an outcome package to Unity Catalog.
4. Adapt to your data — map your existing sources to the canonical core, then build on the outcome packages.

Adapting ORDM to your data

Most retailers don't start from a clean slate. The Vibe Modeling Agent is a companion open-source tool that helps adapt ORDM to your existing data landscape — your naming, your source systems, your custom entities — keeping what fits and customizing what doesn't. See docs/vibe-modeling-integration.md.

Documentation

• Architecture — the layered model in detail
• Data model principles — the design conventions ORDM follows
• Retail outcome map — the 12 outcomes
• Adapting ORDM with the Vibe Modeling Agent
• FAQ

Contributing

ORDM is built collaboratively and we welcome input from the retail community. Open an issue to report a problem or suggest an improvement, or open a pull request to propose changes. Accepted code/content contributions may require signing a Contributor License Agreement (CLA) — we'll guide you through it on your first PR.

License

ORDM is released under the Apache License 2.0 — chosen so partners and customers can freely adopt, extend, and build on the model. See LICENSE.md.

Acknowledgments

ORDM is built partners first — our retail solution partners and domain experts lead the way, with Databricks and the broader retail community building alongside them. Independently authored; not a redistribution of any third-party proprietary data model. See NOTICE.md.