*Reducing E-commerce Compliance Costs by 95%: Building GST Filing for Indian Sellers

Managing seller accounts on Meesho, Amazon, and Flipkart, covering listings, inventory, sales dashboards, and returns, I kept running into the same recurring task: GSTR-1.

Every filing period, each seller has to file GSTR-1, the return that reports their outward sales. For a seller on three marketplaces, that means taking three sales reports in three different formats and reconciling them into a single return. Done manually, it costs roughly Rs. 24,000 and takes hours, and the work grows with the business. Compliance was becoming a bottleneck for exactly the sellers who were scaling.

Seeing this process up close in real seller accounts showed me where the time actually goes. The GSTR-1 form isn't the hard part. The data is:

• Three schemas. Amazon, Flipkart, and Meesho each export sales data in their own format, with their own column names and conventions.
• Returns and adjustments. Returns, cancellations, credit notes, and debit notes all change the taxable totals, and each has to be treated the way the GST portal treats it.
• Precision. Tax values have to be rounded exactly the way the portal calculates them, or the numbers don't reconcile.
• Multiple files. Monthly and quarterly filings mean merging several reports into a single return.

The existing options were spreadsheets or a chartered accountant. Neither gets cheaper or faster as order volume grows.

I designed GST Filing around one principle: a seller should get from marketplace export to a filing-ready return without needing to know how GSTR-1 is structured.

• Bring marketplace data as-is. Amazon, Flipkart, and Meesho sales data goes in without manual reformatting.
• Compute what the portal expects. The platform generates the B2CS, SUPECO, and DOC_ISSUE sections with validation and precision rounding.
• Handle the messy cases. Returns, cancellations, credit notes, and debit notes are applied to match GST portal calculations.
• Merge filing periods. Multiple files can be combined for monthly or quarterly filings.
• Export and understand. Output is available in Excel and JSON, alongside state-wise and platform-wise sales analytics.

GST Filing is built with Next.js, React, TypeScript, and Node.js. At its core is a data pipeline that runs every upload through the same stages:

1. Ingest: parse each marketplace's Excel/CSV sales export.
2. Normalize: map Amazon, Flipkart, and Meesho fields onto a common structure, so every later step is marketplace-agnostic.
3. Adjust: apply returns, cancellations, credit notes, and debit notes the way the GST portal does.
4. Aggregate: build B2CS (grouped by place of supply and tax rate), SUPECO (supplies made through e-commerce operators), and DOC_ISSUE (documents issued), with validation and precision rounding.
5. Merge and export: combine files across periods and generate Excel and JSON output, plus state-wise and platform-wise analytics.

• next.js
• react
• typescript
• node.js
• data analytics
• excel/csv processing

GST Filing is live at gstfiling.app and supports sellers on Amazon, Flipkart, and Meesho. It now has 1.3K+ active users and draws 5K+ visitors to the website every month.

The value is easy to explain because sellers already pay for this work every filing period: the same return, in minutes instead of hours, for a fraction of the cost.

• 1.3K+ active users
• 5K+ monthly website visitors
• 95% lower compliance cost, from Rs. 24,000 to Rs. 1,200
• Report generation cut from hours to under 2 minutes
• Eliminated reliance on manual, CA-assisted processes
• Supporting sellers across Amazon, Flipkart, and Meesho

• Domain experience is the best user research. Managing seller accounts showed me the problem before I tried to solve it, so I was building for a workflow I had seen up close rather than one I had imagined.
• Correctness is the product. For compliance software, a number that doesn't match the GST portal is worth nothing. Matching the portal's rounding and its treatment of returns and credit notes mattered more than any feature.
• The edge cases are the value. Returns, cancellations, and credit and debit notes are exactly what makes manual filing slow. Handling them well is what lets a seller trust the output.