A bilingual merchant can have a working Chinese URL while its safety note still describes yesterday’s English product. Commission a translation change register by field and source revision. This solves content freshness and responsibility, separately from hreflang or URL pairing.
Primary evidence and claim boundaries
Shopify distinguishes translated, outdated and untranslated content; source changes can leave existing translations marked outdated. Primary reference · Read 17 October 2026.
Filled acceptance example
Synthetic teaching scenario, not a client case, provider integration test or business result. Scroll the example table on narrow screens.
| Field / source revision | Chinese review record | Release decision |
|---|---|---|
| P17 care text r4 → r5 | Translation still based on r4 | Hold this field; assign translator |
| P17 title r2 unchanged | Reviewed translation matches r2 | Retain approval; no needless retranslation |
| New P18 returns link r1 | No Chinese content or verified fallback | Untranslated; review actual storefront output |
| P17 care zh revision z6 | Reviewer compared to source r5 | Approve r5/z6 pair, not the whole catalogue |
Practical handover procedure
- Inventory customer-critical fields, including app-controlled text, notifications and policy links. Assign each field a source revision and a named review role.
- When source content changes, invalidate only affected translation approvals. Compare meaning, numbers, exclusions and links; a fluent sentence can still preserve an obsolete fact.
- Preview the actual language/market and app surfaces. Record the output and approve the exact revision pair before the merchant publishes; keep incomplete items visible in the register.
Interpret the outcome
In this example three fields have three different decisions. “Chinese enabled” cannot replace a field-level review. Reusing the title approval saves work safely because its source did not change, while the care note requires human review before a claim of bilingual parity.
Acceptance decision
A handover should identify every blocked field, the source revision it needs and the approved replacement revision. Do not label a progress percentage as editorial approval.
Limits and unverified work
Synthetic revision register. No translation app was run and no store content was changed. Vendor statuses are review cues, not proof of accuracy; app content and jurisdiction-specific policy need their own evidence.
Download the filled bilingual acceptance record — Contains these inputs, steps, expectations and limits. This is a teaching record, not an executable integration.