Customer and task
Help a fictional cross-border retailer handle product content and support replies. Markets use different languages and policy versions; some claims are restricted and some records contain personal contact details. Check whether the content may be used, as well as whether the translation is fluent.
Try it
Review records by market, language, policy version and risk. Plan contact-data handling, human review and error escalation, then explain how to stop using a withdrawn policy.
How to use this data
The practice data contains risk types, not consumer text. Use it to review the workflow, not to establish compliance with a country’s policies or estimate real-market returns.
The dataset contains 20 practice records. Use paper or a spreadsheet to count and compare task groups, then list possible causes and questions to test. No coding is required.
Download synthetic JSON dataSample distribution
View chart data
| human_result | Records |
|---|---|
Approved approved | 6 |
Approved after redaction redacted_then_approved | 4 |
Rejected rejected | 6 |
Rewritten rewritten | 4 |
Field reference
| Field | Meaning and limitation |
|---|---|
id | Synthetic event identifier for cross-referencing. |
market | Target market code, such as US, DE or JP. |
language | Task language code, such as zh, en, de or ja. |
task | Task type, such as listing (product content) or return_reply (reply to a return request). |
policy_version | Market-policy version used for the task; validity needs verification. |
risk | Recorded risk category; none means no risk was labeled in the record. |
contains_contact | Whether contact information is flagged; no actual consumer text is included. |
human_result | Human handling result, such as approved or redacted_then_approved (approved after sensitive information was removed). |
minutes | Synthetic task processing time in minutes; not a direct financial benefit. |
What to write down
Write a one-page analysis: which tasks to address first, supporting records, questions to test, owners and pause conditions. End with a specific next step. Mark missing evidence for follow-up rather than inventing facts.