Data Migration Without Drama: A Practical Sequence
The new system can be perfect and still fail on day one if the data is wrong. A step-by-step sequence for moving data calmly, with trial runs and checks that catch problems early.

It is Saturday night before go-live. The new ERP (enterprise resource planning) system is configured, the team is trained, and the data load has just failed for the third time because some product codes contain spaces and several hundred customers have no tax number. By Sunday morning, people are typing records by hand.
Data migration is usually treated as a technical task at the end of a project. In reality it is a business task that should start near the beginning, and many go-live problems trace back to it.
Decide what to move before deciding how
Before choosing any migration tool, agree which data the new system actually needs. Moving everything feels safe, but it imports old errors, slows the project and clutters the new system with records nobody uses.
Take a building supplies trading company in Amman replacing a fifteen-year-old local accounting package with a modern ERP. A sensible scope might look like this:
| Data | Typical approach | Notes |
|---|---|---|
| Active customers and suppliers | Migrate, after cleaning | Define “active”, for example any transaction in the last two years |
| Products and prices | Migrate, after cleaning | A good moment to retire obsolete items and standardise units |
| Open sales and purchase invoices | Migrate as open items | Each one must match the customer or supplier balance exactly |
| Stock on hand | Migrate quantities and values at cutover | Usually based on a physical count close to go-live |
| General ledger | Opening balances, sometimes monthly totals | Agree with finance and your auditor how much history is needed |
| Closed transactions from past years | Archive, do not migrate | Keep the old system or an export available read-only for reference and audit |
Agreeing this table early prevents months of debate later. It also sets a realistic workload for the cleaning that follows.
Write the agreed scope down and have finance, operations and the project sponsor sign it. Requests to bring across one more year of history tend to appear late in a project, and a signed scope makes it easier to weigh each one on its cost rather than its urgency.
The sequence, step by step
1. Profile the existing data
Extract the data from the old system and look at it honestly. Count records, find blanks in important fields, spot duplicates, and check formats such as phone numbers, tax numbers and units of measure. This takes days, not weeks, and it tells you how much cleaning lies ahead.
2. Map old fields to new ones
For every field in the new system, note where its value comes from, whether it needs transformation, and what to do when it is empty. Old systems often store several things in one field, such as a customer name with the branch appended, which must be split. Business owners should sign off the mapping, not only the technical team.
3. Clean in the source or in a staging file
Decide where cleaning happens. Fixing records in the old system keeps everyone working on the same data, but some changes are impossible there. A staging file, meaning a separate spreadsheet or database where records are corrected before loading, gives more control. Whichever you choose, assign each cleaning task to a named person from the business: sales cleans customers, purchasing cleans suppliers, the warehouse cleans products.
4. Run trial loads
Load the data into a test copy of the new system at least two or three times before go-live. The first trial usually fails in places, which is exactly why you run it. Each run reveals format errors, missing mandatory fields and mapping mistakes while there is still time to fix them calmly.
5. Reconcile every trial
After each load, compare totals between old and new: number of customers, total receivables, total payables, stock quantity and value by warehouse, trial balance by account. Any difference must be explained. Repeated reconciliation is what turns a hopeful migration into a reliable one.
Planning the cutover
Cutover is the short window when the old system stops and the new one starts. It should feel like repeating a rehearsal, not improvising. A typical checklist includes:
- Freeze date and time agreed. After this point no new transactions go into the old system, and everyone knows it in advance.
- Final transactions posted. Last invoices, receipts and stock movements entered and checked in the old system.
- Stock count completed. Physical count finished and variances approved, so opening stock is trustworthy.
- Final extract and load. Using the same scripts or templates as the last successful trial, with no last-minute changes.
- Reconciliation signed off. Finance and operations confirm the opening position before users start working.
- Fallback plan ready. A clear decision point and a way back if something fundamental fails.
Choose the timing carefully. Month-end or year-end cutovers simplify opening balances, but they collide with the busiest period for finance. Many companies in the region also avoid peak seasons such as the weeks before Ramadan or the start of the school year, when order volumes spike.
The first weeks after go-live
Expect a period of small data issues: a customer with the wrong payment terms, a product with the wrong unit, a supplier missing a bank account. Set up a simple log where users report data problems, assign an owner to fix them daily, and track the count going down. Keep the old system available read-only so people can check history without guessing.
This is also the moment to put rules in place that keep the new data clean: mandatory fields, controlled lists, and a clear owner for creating customers and products. Some of these checks can be automated, such as alerts when a record is created without a tax number. Otherwise, within a year the new system can carry the same problems as the old one.
A migration that supports the new system
If your project is already under way, check two things this week: whether the scope table has been agreed and signed, and whether the plan includes at least two full trial loads with reconciliation before go-live. If either is missing, adding it now costs far less than discovering the gap on cutover weekend.
Data architecture and migration are a core part of our Enterprise Systems service. We help you scope, map, clean and reconcile your data, so the new system starts with information your team can trust.



