1. Why now — the 2030 deadline
SAP set the end-of-maintenance for ECC 6.0 (the classic SAP ERP) to end of 2030. Companies that haven't migrated by then run without security patches, updates or official support. For many in regulated industries that's a no-go.
A typical S/4HANA migration takes 18-30 months. If you need to be live by 2030, you should start no later than end of 2027. That makes 2026/2027 the peak years for SAP transformation projects. External consulting capacity is already tightening; by 2028 the industry expects acute shortages.
Three strategic paths exist:
- Brownfield (System Conversion): existing ECC is converted to S/4HANA — all data, customizing and (with adjustments) custom code stay. Lower risk, shorter runtime, but little innovation. ~60 % of projects today.
- Greenfield (New Implementation): S/4HANA is set up completely new. Data is migrated selectively, processes redesigned from scratch. High transformation potential, but double the runtime and cost.
- Bluefield / Selective: hybrid approach — data and processes selectively taken over, critical areas rebuilt. Methodologically demanding, fits multi-site enterprises.
2. Typical structure in 6 phases
Independent of strategy, an S/4HANA migration follows similar phase logic. The SAP methodology is called "SAP Activate" and consists of these phases:
| Phase | Activities | Duration |
|---|---|---|
| 1. Discover | Readiness check, business case, Brownfield vs Greenfield decision | 2-3 months |
| 2. Prepare | Detailed planning, team setup, infrastructure, custom code analysis | 2-4 months |
| 3. Explore | Fit-gap analysis, prototypes, process design (Fiori, Universal Journal) | 3-5 months |
| 4. Realize | Configuration, code migration, data migration, integration tests | 6-12 months |
| 5. Deploy | UAT, cutover plan, go-live, hypercare (4-8 weeks) | 2-4 months |
| 6. Run | Stabilisation, innovation backlog, continuous improvement | ongoing |
In the mid-market (500-3,000 employees), Discover + Prepare + Explore combined should not exceed 6 months. Taking longer here risks momentum loss and budget wasted in the pre-phase.
3. Realistic budget ranges by company size
The question "what does an S/4HANA migration cost?" is the most-asked and hardest to answer. Roughly — based on 2024/2025 industry benchmarks:
| Company size | Brownfield | Greenfield | License share |
|---|---|---|---|
| Small (≤ 500 emp.) | €800k - 2M | €1.5 - 4M | 15-20 % |
| Mid-market (500-3,000) | €2 - 5M | €4 - 12M | 12-18 % |
| Enterprise (3,000+) | €5 - 30M | €15 - 80+M | 8-15 % |
What makes the cost range so wide:
- Custom code volume: each Z-program costs €2,000-8,000 to migrate. 3,000 programs = up to €24M for code alone.
- Interfaces: €15,000-80,000 per interface depending on complexity.
- Data migration and cleanup: 10-25 % of total budget, often underestimated.
- Change management: 8-15 % of total budget, often completely forgotten.
- Hypercare and stabilisation: 10-15 % of total budget, takes longer than planned.
Field tip: always plan a 20-25 % risk reserve on top of the initial budget. A Gartner study (2024) shows: 67 % of all S/4HANA projects exceed their budget by an average of 28 %. Without the reserve, you escalate mid-project.
4. Where AI actually helps today
"AI in the SAP migration" is overhyped in many sales pitches. Realistically AI helps today in three clear areas:
4.1 Readiness analysis and custom code classification
SAP Signavio and tools like Panaya or snap_INSPIRE scan the existing ECC code and automatically classify:
- Which Z-programs are still in use (usage statistics)?
- Which are S/4HANA-compatible, which need adjustment, which must be rebuilt?
- Which standard transactions are simplified or replaced in S/4HANA (Simplification List)?
Time saved: what used to be 12 weeks of manual work now runs in 3 weeks.
4.2 Project plan generation and phase structuring
Tools like PathHub AI or specialised SAP planning AI generate from project context (size, Brownfield/Greenfield, industry) realistic project plans with phases, milestones, risks and budget ranges. Output is never 1:1 usable but saves 60-80 % of initial effort.
4.3 Test automation and defect classification
SAP Tricentis and CrossLogic use ML to identify critical paths from historical test cases and generate automated regression tests. In large data migrations, AI models classify anomalies (plausibility check) and prioritise defects.
Where AI does not help yet (despite the hype)
- Strategic Brownfield/Greenfield decision: stays business-case + consulting
- Process redesign (fit-to-standard): needs functional expertise
- Change management and adoption: stays human
- Cutover planning in complex multi-system landscapes
5. The critical risks — and how to spot them early
- Custom code tsunami. Mid-market ECC systems average 2,000-5,000 Z-programs. 60-70 % are unused. Without early triage you migrate dead code. Fix: usage tracking in Discover phase, clarify with business which programs are still needed.
- Data quality. Duplicate suppliers, outdated material masters, empty mandatory fields — all become problems at migration. Per 100,000 master records expect 2-4 weeks cleanup per data domain. Fix: data cleanup track in parallel to the migration, not after.
- Interface underestimation. Each ECC interface must be retested, often rebuilt. Without complete inventory you find ghost interfaces later. Fix: interface inventory in Discover with IT Operations.
- Business team availability. Key users from business are essential — and day-to-day must continue. Without clear allocation (at least 30 % capacity), adoption fails. Fix: allocation in steering committee as hard constraint.
- Cutover duration. The final cutover (ECC shutdown → S/4HANA up) takes 3-7 days and must fit production schedules. Planning it only at the end leads to escalation. Fix: cutover strategy already in Prepare phase.
- Consultant dependency. External consultants are expensive and gone after the project. Without internal know-how build-up the company stays dependent. Fix: 20-30 % internal team ramp-up as mandatory KPI.
6. Real example: mid-market machinery manufacturer
A machinery company in Baden-Württemberg, 1,200 employees, 3 plants, ECC 6.0 in use since 2008.
Discover (3 months, Q1/2025): Readiness check via SAP Signavio. Result: 2,847 Z-programs, of which 1,640 unused (last 12 months). 47 interfaces, 14 with critical dependencies. Most custom reports replaceable by standard Fiori apps. Decision: Brownfield conversion, budget €3.8M, 22 months runtime, go-live Q3/2027.
Prepare (3 months): Team setup with 14 internals + 18 externals. Cleanup track for material and supplier masters starts in parallel. Cutover strategy: 5-day Pentecost weekend cutover in 2027.
Explore + Realize (12 months): Fit-gap workshops per module (FI, CO, MM, SD, PP). Universal Journal setup without New GL migration (already S/4 ready). Custom code reduced from 2,847 to 487 (83 % less). Data migration in 4 waves.
Deploy (3 months, Q2-Q3/2027): Stress test with 1.2M records. UAT with 80 key users. Cutover at Pentecost weekend: 87 hours downtime, controlled restart with 4 critical bugs (all fixed in hypercare).
Run (ongoing): Hypercare 8 weeks, then normal operations. Performance comparison: report generation 4× faster, monthly period close from 7 to 3 days.
Lessons learned: custom code cleanup could have started before Discover (6 months saved). Change management initially underestimated (escalated to double budget). AI readiness check saved ~4 person-months in analysis.
7. Pre-project checklist
Before signing the first contract:
- ✓ ECC end-of-maintenance dates clear: by when is migration mandatory?
- ✓ Brownfield vs Greenfield strategy aligned with the board?
- ✓ Internal team identified and allocated (at least 30 % capacity)?
- ✓ External consulting budget realistic (50-70 % of total budget)?
- ✓ 20-25 % risk reserve above the initial budget?
- ✓ Cutover window roughly aligned with operations?
- ✓ Data cleanup track planned as separate workstream?
- ✓ Change management with its own budget (8-15 %)?
- ✓ Steering committee with clearly defined escalation paths?
- ✓ Internal know-how build-up defined as mandatory KPI?
If you say "no" to more than 3 of these, the project is not ready to start. Better 3 extra months in Discover than a €3M mid-project tear-down in Realize.