The typical starting point
Each month, staff download the GSTR-2B JSON, open the Tally purchase register and match invoices one by one. Mismatches go into a separate sheet, a partner reviews them and the client is called. Multiply that by every client and a large share of the team's month goes into routine matching.
How I would build it
I would map every step of the manual process and rebuild it as a Power Query pipeline. The GSTR-2B JSON loads via From File → JSON; the Tally purchase register loads From Folder. The two are merged on GSTIN + invoice number, with tolerance rules for rounding. Mismatches are flagged with conditional formatting, and each client gets a one-page reconciliation with a match/mismatch status and an unmatched-invoice list. A practice-wide tracker rolls up status for every client.
What it is designed to deliver
A build like this turns matching into a refresh, so staff time goes into resolving the flagged mismatches instead of finding them. The tracker shows partners which clients are done and which are pending before the filing deadline.
