1. Write down what moves
Start with objects and fields, not an app pair. If the ERP owns a sales price and Shopify owns a product description, the integration should make that boundary explicit. Two systems writing the same field need a conflict policy.
A field-ownership table can be short. The important part is making the decisions visible before anyone configures a connection.
| Example requirement | Decision to make | Test to keep |
|---|---|---|
| Product title and price | Which system is authoritative? | Change the source and verify only the owned fields change. |
| Existing product | How is a stable source ID matched to the destination? | Replay the same source and check that the intended product is reused. |
| Inventory | Which locations, units and stock events are in scope? | Check independently; a product recipe does not imply stock support. |
| Order or invoice | When is it eligible, and how are exceptions handled? | Include shipping, discounts, refunds and more than one tax rate. |
2. Check access and the first-import boundary
Choose a first-import boundary: all eligible history, a start date or future changes. Shopify order access is limited to the recent 60 days by default; older orders need approved read_all_orders access in addition to an appropriate order scope.
Check your ERP version, hosting plan and API permissions. Odoo’s documentation places external API access on its Custom plan for Odoo Online. Integrator’s current Odoo connector targets versions 16 to 18 using JSON-RPC; JSON-2 support is separate work.
3. Use a concrete scope example
Integrator’s beta Odoo-to-Shopify recipe maps title, SKU and sales price. Each Odoo variant becomes one Shopify product. An optional Odoo boolean flag can narrow the product set.
This recipe leaves Shopify tags, collections and descriptions alone. It does not synchronize stock, images or Shopify orders into Odoo. If those are requirements, they need a separately assessed flow rather than an assumption that they come with the connection.
Odoo and Shopify are both supported in beta. Check the selected recipe’s fields and a dry run before enabling it.
4. Prove the result before scheduling
Run a preview and dry run first, then approve a small live test in an appropriate test environment. Check both the record history and the destination itself.
- Use a product with an existing destination record.
- Test a missing SKU and a source value that should be filtered out.
- Change a price and verify the intended product updates.
- Replay the same record and inspect its destination identity.
- Confirm untouched fields remain intact.
- Assign a person to review the first scheduled runs and handle exceptions.
