The client is a machinery manufacturing company with ~50M RMB annual revenue and 80+ employees. When they came to us, their entire business was managed through 20+ Excel spreadsheets — one for purchasing, one for inventory, one for sales, one for finance — all manually linked by copy-paste, often mismatched.
Client Pain Points
- Purchasing, inventory, and sales data scattered across different Excel files; monthly reconciliation took 3 days
- Boss wanting a sales report meant finance manually summarizing for half a day
- Order progress was all word-of-mouth; no one could tell which stage an order was at
- Employee handover was difficult; only the creator understood their Excel logic
Requirement Research (Weeks 1-3)
We did not jump into system design. Instead, we spent 2 weeks walking through the client business processes — from purchase orders to warehouse receipt, from sales orders to shipping and payment. We collected every Excel sheet used in each step.
We discovered a key issue: some of the client business processes were irrational (e.g., missing quality checks, verbal purchase approvals). Directly moving existing processes to the system would solidify the problems. We proposed process optimization suggestions, adjusting 3 workflows after discussion with management.
System Design (Weeks 4-6)
Modules: Purchasing, Inventory, Sales, Production, Finance, Reporting Dashboard, Permission Management.
Design principle: First solve data interconnection (core), then optimize processes (advanced), finally add data analytics (value-add). Each module was confirmed with department heads for workflows and field requirements.
Development (Weeks 7-14)
Tech stack: Laravel backend + Vue3 frontend + MySQL.
Strategy: Module-by-module iteration, delivering a usable module every 2 weeks for client trial. This approach lets clients see results early and adjust direction, rather than discovering misalignment at the end.
The biggest challenge was data migration — 20+ Excel sheets with inconsistent formats, merged cells, and manual annotations. We wrote a data cleaning script to import Excel data into intermediate tables for cleaning and validation before loading into the production system. This process took a full week.
Launch (Weeks 15-16)
Phased training: First train department key users, who then train their colleagues. More effective than us training everyone directly, as key users understand department-specific use cases better.
Parallel operation: System and Excel ran in parallel for 2 weeks, comparing data consistency. Official switch after verification.
Post-Launch Changes
- Monthly reconciliation from 3 days to half a day (auto-summarized by system)
- Boss can view real-time sales reports and inventory data anytime
- Order progress visible in real-time, no more phone calls
- New employee training from 2 weeks to 3 days (system is easier than Excel logic)
- Data no longer depends on individuals; smooth handover when employees leave
About the Cost
Total investment was ~150,000 RMB. The client initially felt it was expensive, but after calculating: duplicate purchasing from inventory mismatch cost ~80,000 RMB/year; plus reconciliation labor and decision delay costs — the system paid for itself in 6 months.
Final Thoughts
An ERP system does not need to be huge — the key is solving your actual problems. If your business is also suffering from Excel management, contact us — we help evaluate which processes most need systematization, without requiring a big system upfront.