Sales reconciliation and integration operations between a sports brand's Korean online store and its ERP

Period 2026.04 — Present

I handle month‑end sales reconciliation and operate the integration server between the two systems. Among the differences the tool finds, I correct sales and settlements that were actually missing.

Reconciliation tool that compares about 390,000 rows from 58 days of records and sorts differences by cause

Team: Lukuku operations teamReconciliation and integration server
Role: Sales reconciliationReconciliation tool developmentIntegration server operationsSettlement corrections

PythonShopifyMicroservices

Key‑by‑key matching of orders, shipments, returns, exchanges, coupons and points
Settlement correction process organized by how exchange orders lose settlement data
Voucher prices pulled from the ERP to calculate discounts per order item

Real gaps versus counting differences

Problem: The two systems record data differently and at different times, so spreadsheet comparisons mixed real gaps with counting differences.
Solution: Sorted differences into ERP‑only, store‑only and amount mismatches, and added a step that queries both source records again for each difference and attaches a cause. After the partner said the counts looked inflated, I refined the rules separating analysis errors from real differences. Narrowed the records to 16,365 differing rows. The 10,203 rows the partner needed to check and correct went out as 49 files by category.

Timing differences in month‑end missing‑sales lists

Problem: Each month‑end list from the partner included sales that had already been sent after the list was generated.
Solution: Checked each item against the store's send history to identify the cause. I left already sent items out to prevent duplicates. Resent the missing or partially sent sales data according to ERP rules.

Sales data after a size change

Problem: For orders shipped in a different size, the change record, shipping label, shipping status and shipment integration all carried the new size. But the sales data alone reached the ERP with the original size.
Solution: Narrowed the cause to the product‑change integration logic. I resent the sales data to the ERP and then moved on to fixing the logic. Finished resending the affected sales data.

Monthly settlement checks

Problem: Each month the store's orders had to match the ERP settlement. Resending orders with broken settlement data as they were could send duplicates.
Solution: For September, I listed 41 orders with broken settlement data and the records missing master data. For October, I re‑extracted 72 rows and sent them all except 2 that had to be excluded. I held back further resends that risked duplicates. Finished the September and October reconciliations.