Export Docs Migration That Won't Break a Live Shipment

Most Indian export businesses do not have a documentation system. They have a working arrangement: a commercial invoice made in Excel or Tally, a packing list edited by hand, a CHA who gets documents over email and WhatsApp, and a CA who reconciles everything at quarter end. It holds together because two or three people know the sequence by heart. That is also why nobody wants to touch it.

The fear is reasonable. A shipping bill filed with the wrong HS code, an e-BRC that never matches an invoice, a GSTR-1 Table 6A that disagrees with your shipping bills — these are expensive to unwind. So the migration has to be phased, and it has to assume a shipment is already in flight on day one.

Map the workflow before you move any of it

Before choosing anything, write down what actually happens on a shipment. Not the ideal process — the real one. For most SMBs exporting 20-80 consignments a month, that means inventorying:

• Every document produced, by whom, in which tool (proforma invoice, commercial invoice, packing list, shipping bill inputs, BL/AWB instructions, Certificate of Origin, LUT copy, bank realisation advice)
• Who consumes each one downstream — CHA, freight forwarder, your bank's trade finance desk, DGFT, your own CA
• Which fields on those documents are typed by hand and which are copied
• Where the same value gets entered twice (usually consignee address, HS code, invoice value, and Incoterms)
• Which documents must exist before the shipment moves, and which are produced after

That last split matters most. Pre-shipment documents block a container. Post-shipment documents — e-BRC, RoDTEP claims, GST reporting — only block money. You can migrate the slow, forgiving half first.

The three-phase path

Phase 1: shadow mode. Run the new system alongside the old one for a fixed window — typically 6-10 shipments, or one full month, whichever comes first. The new system generates every document, but nothing goes to your CHA from it. You compare side by side and log every difference. Budget for 5-15 mismatches per shipment in the first week; most are formatting, a few are real. Expect 2-3 hours a day for one person — usually the ops lead who already knows the sequence, not a developer.

Phase 2: cutover by lane, not by date. Do not flip the whole company on a Monday. Cut over one slice at a time: one product category, one port pair, one branch, or one incentive scheme. A realistic sequence for a ₹15-40 crore exporter is a single HS chapter for three weeks, then a second, then the rest. Avoid cutover in the last week of a quarter or during your peak shipping season — month-end already carries GST and realisation work, and there is no reason to stack.

Phase 3: freeze and decommission. The old spreadsheet goes read-only, not deleted. You will want it for the audit trail on shipments filed before cutover, and for the 12-month lookback on realisation.

A worked example

Say you ship 40 consignments a month, roughly 10 a week, across two CHAs. A single unwound mistake — an invoice value mismatch that holds up an e-BRC, for instance — can take 3-6 weeks and several calls to your bank's trade desk to resolve.

Attempt a big-bang migration in week 1 and hit a mismatch rate of even 10%, and that is four shipments a month needing manual repair, in the middle of your operating rhythm. Shadow mode costs the same setup effort but moves the repair work to an internal comparison instead of a live consignment. In our experience, teams that dual-run for a month land under 2% mismatch on cutover day. Teams that skip it land wherever their data quality already was — which is usually the problem they were trying to solve.

Where the data model fights back

Three things break reliably, and they are worth designing for before anyone writes code:

• Currency and rounding. Export invoices are raised in USD or EUR with a rupee value derived from an exchange rate on a specific date. If your system stores one and derives the other at display time, GST reporting will drift from the invoice the buyer and bank hold.
• Invoice numbering. Many SMBs run separate series per branch, per financial year, or per buyer. Changing the numbering scheme mid-year creates a reconciliation problem with your bank and your CA that is not worth the tidiness.
• Document versions. Shipping bills get amended — quantity, value, HS code. Store the amendment history rather than overwriting it, because that history is what you show during scrutiny.

Keep the old document format as an export option for at least two quarters. Your CHA will ask for it, and saying yes is cheaper than retraining them.

Where to start

Pick one shipment this month. Rebuild its documents in the new system, send them nowhere, and compare field by field against what actually went out. That single exercise tells you more about your data quality than any vendor demo — including ours. We build export and operations workflows for Indian SMBs at Maysan Labs, and the first thing we usually do with a client is exactly this: one shipment, side by side, before anything gets switched over.