Every factory we've digitized ran on the same stack: 6–15 Excel sheets, two WhatsApp groups, and one person named "Sharma ji" who alone knows how the pricing sheet works. Moving that to an ERP without stopping production is a process problem more than a software problem. This is the playbook we actually use.
Step 1 — Audit the spreadsheet jungle (Week 1)
List every sheet, who fills it, who reads it, and when. The goal isn't to copy Excel into software — it's to find out which sheets are the real system (usually 4–5) and which are workarounds for the last failed software. The workarounds tell you what the ERP must not repeat.
Step 2 — Map fields, kill duplicates (Week 1–2)
The same item is "MS Pipe 2inch", "MS PIPE 2"", and "pipe-2" in three sheets. Before any code: one item master, one customer master, one unit convention. This master-data exercise is 30% of migration success and costs nothing but discipline.
Step 3 — Clean and load data (Week 2–3)
- Export everything to CSV; de-duplicate programmatically, verify manually with the owner.
- Load opening balances, not full history. Ten years of old entries add noise, not value — keep old Excel as archive.
- Freeze a cutover date for stock counts so opening stock is real, not "approximately".
Step 4 — Parallel run (2 weeks, non-negotiable)
Run Excel AND the ERP together on live data for two weeks. Daily 10-minute mismatch review: every difference is either a data error (fix the data) or a workflow gap (fix the software). Skipping the parallel run is the #1 cause of "ERP fail ho gaya" stories.
Step 5 — Train by role, not by feature tour (Week 4–5)
- Store keeper learns 3 screens. Accountant learns 4. Owner learns the dashboard. Nobody sits through a 4-hour "full training".
- Hindi/Hinglish cheat-sheets with screenshots at every desk beat any manual.
- Golden rule we enforce: if it's not in the ERP, it didn't happen. Management must refuse Excel reports after cutover — adoption follows leadership, always.
Step 6 — Cutover and the first month
Go live on a Monday after a weekend stock count. Expect a slow first week — speed returns by week three and then beats Excel permanently. Keep a daily 15-minute support window with your vendor for month one (we do this on WhatsApp), then weekly.
What this delivers
Our manufacturing clients typically see 30–40% less admin time, same-day GST billing instead of month-end marathons, and — the owner favourite — live stock and outstanding figures on their phone instead of evening phone calls to three departments.
FAQ
How long does the whole migration take?
5–7 weeks alongside ERP development's final phase. The ERP itself takes 3–4 months; migration overlaps the last stretch.
Will production stop at any point?
No. The parallel-run method exists precisely so the factory never depends on an unproven system.