Data analysis
Reports and exports
Request complete Excel, CSV, and scheduled reports with traceable detail and privacy boundaries.
#Define the report before exporting
AI can present findings and tables in a conversation or generate Excel and CSV when requested. Exporting delivers an analysis result; it does not require a dedicated product page with the same report name.
Specify the Shop or account, period, date basis, filters, row meaning, required columns, and whether raw detail is needed.
Export all order-SKU lines for Brightloom US last month using Shop-local payment dates, as Excel. Preserve order, product, and SKU identifiers as text. Separate currencies and include a summary, date definitions, and data gaps. Export every matching row, not just the preview.
#Choose a useful output
| Need | Suggested output |
|---|---|
| Quick exception check | Short findings, key figures, and next checks |
| Operating review | Summary, comparison periods, explanations, and supporting detail |
| Human follow-up | Creator, product, or order list with owners and decisions needed |
| Finance reconciliation | Raw settled income, fees, adjustments, payout links, and exceptions |
| Further processing | Complete Excel / CSV with field definitions, period, and filters |
A small detail preview is not a complete export. Explicitly request all matching data and ask for row counts and coverage. Source limits, unready data, or incomplete history must be stated rather than hidden behind a “complete report” label.
#Keep detail and totals separate
A workbook can contain several sheets, but define each sheet's row grain: an order, order SKU, aftersales request, Sample Application, or payout.
- Preserve long order, Creator, and SKU identifiers as text to avoid rounding or scientific notation.
- Separate raw records from derived findings, keeping identifiers and units for reconciliation.
- Keep currencies separate and label rates versus multiples.
- Do not repeat request totals across product lines or shared warehouses across Shops.
- Leave unavailable data blank; do not silently fill zeros or infer anonymous buyer identities.
Finance exports can retain raw fields for reconciliation. If your company needs an existing template, upload it and explicitly request column mapping, sheets, and formatting. Template labels do not redefine source fields. Preserve financial adjustments without SKUs.
#Make cross-business reports auditable
Create a product review with separate sheets for 30-day sales, ads, Affiliate, returns, and current stock. Join by verifiable product or SKU relationships and list unmatched items. Do not add advertising attribution, Affiliate attribution, and Shop GMV together, or label current stock as 30-day inventory history.
Sources do not always match one-to-one. AI should explain matching rules, duplicates, and unmatched coverage rather than discard exception rows to produce a tidy table. See Interpreting results.
#Daily and weekly reporting
Validate the report in Chat before requesting a Scheduled Task. For example:
Every Monday at 09:00 Asia/Shanghai, prepare a weekly report for these three Shops covering the previous complete calendar week. Use each Shop's local payment dates, separate currencies, and timestamp current inventory. Show the schedule and delivery destination first; wait for confirmation before creating it.
Define the execution time zone, relative dates, target Shops, format, destination, and failure notification. Test once to confirm the file is readable, recipients are correct, and the same report is not delivered twice. The handling Desktop must remain online.
#Files and privacy
Exports retain account and data-access restrictions. Keep bank information masked and share buyer, Creator, and team details only with authorized recipients. Creating a file does not authorize external forwarding. Confirm the audience before sending it to a group.