
ERP Data Migration Checklist for Singapore SMEs: What to Migrate and How to Validate It
ERP data migration goes wrong for four predictable reasons: the data volume is underestimated, the source data is poor, too much history is moved, and the result is not validated before go-live. The fix is a plan with named owners, a trial migration and a signed-off validation checklist. This guide sets out that plan for Singapore SMEs moving from spreadsheets or accounting software into an ERP such as Odoo.
Why is data migration the riskiest part of an ERP project?
Migration makes every existing data problem visible and moves it into a system that everyone will trust. Duplicate customers, products with no cost price, stock that exists only on paper and transactions posted to the wrong account all carry over unless someone fixes them first. A wrong opening balance or phantom inventory then affects reports for years. Our guide to why ERP rollouts fail at the data layer covers the pattern in more depth.
What are the four common migration failures?
| Failure | What it looks like | What prevents it |
|---|---|---|
| Underestimated volume and complexity | Data sits in five places: the accounting system, price lists in a spreadsheet, contacts in a CRM, old purchase orders in a retired system, product specifications in shared folders | A full inventory of every data source, including spreadsheets and folders nobody mentioned at scoping |
| Poor source data | Duplicates, missing cost prices, stock that does not exist, incomplete supplier records | Cleansing before migration, with a named owner for each data set |
| Migrating everything | Years of history and every customer who ever ordered, adding cost and risk with little benefit | Migrate current masters, open items, balances and a defined history window; archive the rest |
| Weak validation | Errors found by users after go-live | A documented checklist with owners and sign-off before the system opens |
What should you migrate, and what should you leave behind?
Migrate what the business needs to operate on day one and to report across a sensible period. Archive everything else in a place that remains searchable.
| Data set | Recommended approach | Check before sign-off |
|---|---|---|
| Customer and supplier masters | Active records only, deduplicated | Count matches the cleansed list; sample of 20 records checked for tax and payment terms |
| Product masters | Active items with costs, units and categories | No item without a unit of measure or cost; sample checked against price lists |
| Open sales and purchase orders | All outstanding orders | Every open order traced to a source document |
| Open invoices and bills | All unpaid items with due dates | Totals agree to the receivables and payables ledgers |
| Stock quantities | Counted positions at the cutover date | Totals agree to a physical count |
| Financial opening balances | Closing trial balance of the old system | Signed off by finance against final accounts |
| Transaction history | A defined window, commonly 12 to 24 months, for reporting | Report totals agree to the old system for the same period |
What is the step-by-step migration plan?
- Inventory the data (start in week 1). List every source, its owner, its format and its quality. Include spreadsheets on personal laptops and old backups.
- Set a quality standard. Define what clean means for each data set, for example one record per customer, a unit of measure on every product, a GST status on every supplier.
- Cleanse at the source. Business users fix the data. Do not plan to fix it during loading, because errors loaded into the ERP become errors in every report.
- Design the migration. Map each source field to an ERP field and document each transformation, such as merging name fields or converting units.
- Run a trial migration. Load the cleansed data into a test system. It exposes mapping mistakes and volume problems that only appear at scale. Fix them, then repeat if needed.
- Run the live migration. Use the latest data at the agreed cutover, then complete the validation checklist before users enter the system.
- Sign off. Business owners confirm each check in writing.
What breaks if migration is rushed: opening balances that do not agree to the audited accounts, stock that differs from the warehouse, customers who receive statements for invoices they already paid, and a finance team that no longer trusts the new reports. Fixing these after go-live costs more than the cleansing would have.
Who is responsible for what?
The implementation partner owns the migration process: tooling, mapping and execution. The business owns data quality: cleansing the source data, validating the loaded data and signing off. Write this split into the project agreement. Disputes usually begin when a data quality problem is treated as the partner’s fault.
What extra care does Singapore data need?
- GST and tax fields. Check GST registration status and the tax treatment on suppliers, customers and products before loading.
- UEN and Peppol details. If you will send e-invoices, capture the UEN and the correct customer identifiers. See our InvoiceNow guide.
- Personal data. Customer and employee records are personal data. Keep migration files in a restricted location, remove copies after go-live and apply the same protection and retention rules as in your live systems.
- Cutover date. Cut over at a month-end or GST period start so that opening balances are clean and returns do not span two systems.
What the ERP cannot own: deciding which records are real, which stock is physically present and which balances are correct. Those are business facts. The system stores and protects them but cannot create them.
Frequently asked questions
How long does ERP data migration take?
The length depends on scope and complexity: the number of records, the state of the data and how quickly your team cleanses it. Cleansing should start in the first week of the project, before the migration phase begins. Messy customer or product data is the usual reason it takes longer.
How much historical data should we migrate?
Migrate current masters, all open transactions, current stock and balances, plus a defined history window, commonly 12 to 24 months, for reporting. Archive older data where it can still be searched.
Can we migrate directly from Excel or accounting software?
Usually yes, because most ERPs including Odoo can import structured files. The risk is not the file format but the content: duplicates, blanks and inconsistent values. Cleanse the data first, then map it to the new fields.
What is a trial migration and is it necessary?
A trial migration loads cleansed data into a test system before the live run. It is necessary because mapping errors and volume problems only show up at scale, and finding them in a trial costs far less than finding them after go-live.
Who signs off the migrated data?
The business owners of each data set: finance for balances and open items, operations for stock, sales for customers and prices. The partner prepares the evidence, and the owners confirm it in writing.
A question for you: if you had to prove today that your customer list, stock count and trial balance are correct, how long would it take?
By Willie Loo, Managing Director, Mxgsoft Pte Ltd and Moxogo Pte Ltd, Singapore. Last reviewed 10 November 2026.
Next step: list every place your operational data lives today and who owns each one. If you want a second opinion, talk to us about a Resource Master Audit.


