Reconciliation
Systems that make every source agree about the same stock and the same money.
| Order | Storefront | Payments | Courier | Warehouse | Result |
|---|---|---|---|---|---|
| #4471 | ₹2,140 | ₹2,140 | Delivered | Shipped | |
| #4472 | ₹890 | ₹890 | Delivered | Shipped | |
| #4473 | ₹1,290 | ₹1,240 | Delivered | Shipped | |
| #4474 | ₹3,600 | ₹3,600 | Delivered | Shipped | |
| #4475 | ₹1,150 | ₹1,150 | Returned | Shipped | |
| #4476 | ₹760 | — | In transit | Shipped |
Checking 1 of 6
- 3 wk → 20 min
- to reconcile stock across every platform
- 1,00,000+
- units reconciled each week without manual work
- 67%
- less manual finance effort on commission runs
Five systems, five truths
Stock lived in an ERP, an order management system, Amazon, Flipkart, and dark stores. Money lived in storefront, payment gateway, courier, and warehouse exports. Each had its own version of events, and the result was overselling, stockouts, and a finance team that spent its month lining up spreadsheets.
Formats as data, not code
The readers for each source sit on file and column configurations stored as data. When a provider renames a column or ships a new export format, the fix is a config change, not a deploy. Retention filters by month and year keep the working set small enough to search and sort quickly.
Leave people the exceptions
Matching is the easy part. The design work is in what happens to the rows that do not match: they get flagged with the reason, the moment they appear, so finance starts from the problem instead of from the spreadsheet.
Reconciliation went from three weeks to twenty minutes, across more than 1,00,000 units a week. On the commission side, refunds adjust payouts automatically with a full audit trail, and manual finance effort dropped by 67 percent while volume grew tenfold.