Connect product records to the catalog with a traceable publishing process.
Copying information from a business system into a website makes version ownership difficult. ImbaHub can scope a data integration between one product source and your public catalog. We establish access, field ownership and publishing rules before promising a connection; an existing ERP or PIM connector is not assumed.
- Source
The product owner controls business facts and validity.
- Mapping
The integration checks identity and agreed fields.
- Publication
The website displays an approved catalog version.
Fetched, approved and visible are different checkpoints.
The import succeeded but the page still shows yesterday's value. Complete?
Not on the import result alone. Check publication, caching and the version shown on the actual product page.
Decide which source owns each field
Article identifiers, pack quantities, documents and editorial descriptions may have different owners. We map imported fields and fields that remain under website editorial control. An empty export value must not erase an approved local description unless that is the agreed rule.
In a fictional trade catalog, the business system owns article IDs and pack quantities while editors write buyer-facing explanations. A discontinued item requires a publishing decision. Its absence from a failed import is not sufficient evidence that it has been discontinued.
Scope mapping and update behavior
We need representative test records, interface documentation and an owner for access. The source may provide a full export or a stream of changes. Stable identity should survive a product rename; category codes and units need documented meanings.
An initial delivery can cover one product family, one source, one language and one publishing route. Scheduled fetching is not automatically real time. Nor does a recent fetch prove that a product specialist recently reviewed the underlying facts.
| Element | Decision | Acceptance evidence |
|---|---|---|
| Identity | Stable article ID and variant relationship | A rename updates the intended record |
| Field ownership | Fields the integration may change | Editorial content survives as agreed |
| Update model | Full export or incremental changes | Missing portions are detected |
| Publication | When imported records become public | Reference items show the approved version |
| Failure | Retain, stop or hide under agreed rules | An owner can find and act on the exception |
Validate before changing the public catalog
A schema can check required fields and types. It cannot alone establish whether a pack quantity is commercially correct. We combine structural validation with agreed reference products and comparison against the authoritative source.
Test an empty response, interrupted pagination, an unknown code, a rename and an intentional withdrawal. Include the publishing step and cache: an imported record does not establish what visitors currently see. Untested source fields remain documented dependencies.
Hand over an observable integration
The handover identifies the source, mapping, import version, logs and recovery procedure. The operator must distinguish the last successful fetch, the latest approved source data and the latest public version.
Account-specific prices, stock, ordering and writing back to the business system are separate capabilities unless explicitly scoped. We propose a tailored scope and price after reviewing the actual interface. Start with one product family and the fields that currently become stale.
Sources and further reading
Questions, answered
Can every business system connect directly?
That depends on its interface, permissions, data and licensing. We review the actual conditions before committing to a connection.
Does a successful sync guarantee current product information?
No. A working integration can transport old or incorrect source data. Freshness and business ownership require their own checks.