Lesson 1 of 5
Why migrations fail
8 min
Listen to this lesson
A new SAP S/4HANA system is useless on day one without data. Customers, suppliers, products, prices, open orders, stock and open balances all have to arrive, correctly and completely, in the short window between shutting down the old system and opening the new one. That is data migration. It is one of the riskiest parts of any SAP program, and one of the most underestimated.
In this course, Contoso Motor Co. has just converted its SAP ECC 6.0 system to SAP S/4HANA. That system conversion kept Contoso's own data in place, so it needed no migration of the kind this course teaches. Right after it, Contoso acquired Alternadores del Norte S.A. de C.V. in Monterrey, Mexico, and now has to load that company's data from its non-SAP legacy ERP into the converted system as new company code 1910 (MXN) with plant 1910 Monterrey. That Wave 2 load is our running example. Luis, the cutover manager, owns the load weekend; Priya, the test lead, owns proving that the migrated data is right; Dan (finance) and Maria (order to cash) own the business data in their areas.
The usual causes
| Cause | What it looks like | Alternadores del Norte example |
|---|---|---|
| Poor source data quality | Duplicates, missing fields, obsolete records, inconsistent formats | Three legacy records for Flotillas Monterrey S.A. de C.V. with different addresses |
| Unclear ownership | Nobody in the business decides what is right, so IT guesses | No one owns supplier bank details; IT copies whatever is there |
| Late testing | The first full load happens weeks before go-live | Open receivables loaded in full for the first time at dress rehearsal |
| Underestimated transformation | New data model or new business rules not reflected in mappings | Legacy customers and suppliers become business partners; new required fields appear |
| Scope creep | Pressure to migrate years of history "just in case" | Request to migrate ten years of closed sales orders |
| Late design decisions | Mapping rules change in the last weeks, invalidating earlier tests | Product numbering changed two weeks before cutover |
Why SAP data migrations go wrong.
Notice that most causes are not technical. Tools load data reliably; the hard part is deciding what the data should be and proving that it is right. That is why this course spends as much time on ownership and testing as on tools.
The migration lifecycle
Every migration follows roughly the same lifecycle, whatever the tool. In SAP's implementation methodology, SAP Activate, data migration work runs through the Explore, Realize and Deploy phases: design in Explore, build and repeated test loads in Realize, and the production load during cutover in Deploy.
- Analyze and scopeWhich objects, how much data, which history, which quality problems
- DesignMapping and transformation rules, cleansing rules, load sequence, owners
- BuildExtraction, transformation and load set up; migration project configured
- Test (mock loads)Repeated rehearsals with validation and reconciliation
- Final load (cutover)Production load in the cutover window, timed and sequenced
- Validate and sign offBusiness validation, reconciliation, sign-off by data owners
What is usually not migrated
A new implementation, or a load of a newly acquired company into an existing SAP S/4HANA system such as Alternadores del Norte's, normally migrates master data (business partners, products, bills of material, prices) and open items (open orders, stock, open receivables and payables, balances), not closed history. Old invoices and closed orders stay in the legacy system or an archive for reference. Organizational structure and configuration (company code 1910, plant 1910 Monterrey, document types) are not migrated as data at all: they are built in the target system and moved by transport.
| Category | Alternadores del Norte examples | How it reaches SAP S/4HANA |
|---|---|---|
| Master data | Business partners such as Flotillas Monterrey S.A. de C.V. and local suppliers, product ADN-200, bills of material, prices | Migrated as data |
| Open items | Open orders, stock in plant 1910 Monterrey, open receivables and payables, balances in MXN | Migrated as data, usually last |
| Closed history | Old invoices and closed sales orders in the legacy ERP | Not migrated; kept in the legacy system or an archive |
| Organization and configuration | Company code 1910, plant 1910 Monterrey, document types | Not migrated; built in the target system and transported |
What would you do?
Six weeks into the Alternadores del Norte load project, the migration specialist reports that 18% of its legacy supplier records lack a valid tax number. The project manager asks IT to "just fill in a placeholder so the load works". What is the best response?
Treat migration as a workstream, not a task
Migration is often planned as a technical task for the last month. In practice it is a workstream that runs most of the program, with its own plan, owners, test cycles, defects and sign-offs. A simple sizing early on helps everyone take it seriously: list the migration objects, estimate record counts per object (for example 1,946 in-scope customers out of 3,180 in the legacy ERP, 1,200 suppliers, 9,000 products, 2,500 open sales order items), name an owner for each and mark the ones with known quality problems. That list becomes the backbone of the migration plan and of the test plan.
- List objectsCustomers, suppliers, products, open orders, stock, open items
- Estimate volumes1,946 in-scope customers, 1,200 suppliers, 9,000 products, 2,500 open order items
- Name ownersA business owner per object, such as Maria for customers
- Flag quality risksMark objects with known problems, like suppliers missing tax numbers
- PlanThe list becomes the backbone of the migration and test plans
- Most migration failures come from data quality, ownership, late testing and underestimated transformation, not from tools.
- Lifecycle: analyze, design, build, test with mock loads, final load, validate and sign off.
- New implementations migrate master data and open items, not closed history; configuration is built and transported.
- Start data cleansing early, owned by the business.
Check your understanding · not graded
Lesson check
1. Which is the most common root cause of migration failure?
2. What is usually migrated in a new implementation?
3. How do company code and plant definitions reach the new system?
4. In SAP Activate, when are repeated test loads typically run?