The three S/4HANA transition paths, the three target editions (Public, Private, on-premise) and how the project approach differs. This article sets out which combinations SAP supports, what each path involves technically, and what each target means for AI and other innovation.
Analysis | ~20 min read | By J. Torre
In brief
- SAP defines three transition paths: new implementation (greenfield), system conversion (brownfield) and selective data transition (SDT, sometimes called “bluefield”). The target edition limits the choice: SAP Cloud ERP Public Edition supports new implementation only; Private Edition and on-premise support all three.
- The paths differ in what they carry over. A conversion keeps your system, data, history and custom code and fixes what the simplification checks flag. A new implementation starts from a clean system and migrates selected data. SDT sits between them: it moves a chosen scope of data and configuration to a new system, or converts a copy of the existing one.
- A conversion has hard technical prerequisites: Unicode, a single ABAP stack, ECC 6.0 EHP0–8 as the start release (with support package restrictions), and a completed Business Partner synchronization (CVI) in ECC before the technical conversion. All custom code and modifications come along and must be scoped, analyzed with ATC and adapted.
- AI access depends on the target. Public Edition has the lightest Joule setup; Private Edition needs more customer-side configuration and, as reported, includes up to three Joule assistants at no extra cost; on-premise and ECC customers can use SAP’s Joule offer only as part of a cloud commitment, on terms that press reports describe differently.
- Mainstream maintenance for SAP Business Suite 7 (ECC) ends at the end of 2027. SAP offers extended maintenance for 2028–2030 at a 2 percentage-point surcharge on maintenance fees, and customer-specific maintenance after that.
- Our recommendation: choose the target edition first, because it narrows the paths; then choose the path by how much of your current landscape you want to keep.
Two decisions, not one
“Greenfield versus brownfield” is two decisions, not one. The first is the target: SAP Cloud ERP Public Edition, SAP Cloud ERP Private, or SAP S/4HANA on-premise (installed and operated by you or a hosting partner). The second is the path: how you get from ECC, or from an earlier S/4HANA release, to that target.
The two are linked. Public Edition is a multi-tenant, standardized service and cannot be reached by converting an existing system. Private and on-premise run the same S/4HANA core as a single-tenant or self-managed system and can be reached by any path. The target therefore comes first.
The three paths
New implementation (greenfield). You install a new S/4HANA system and configure it from scratch, usually following SAP Activate and a fit-to-standard approach. Master data and a selected set of open items and balances are migrated; transaction history is left behind. For customers who already run SAP, this is often called “semi-greenfield” because there is a legacy system to extract from, but the target is built new. Custom code is rebuilt selectively, not carried over.
System conversion (brownfield). You convert the existing system in place. The database moves to SAP HANA (if not already there), the application is upgraded to S/4HANA, and the data model is converted. The Software Update Manager (SUM) with Database Migration Option (DMO) performs this as one technical process, with variants such as DMO with System Move when the infrastructure changes at the same time. Processes, configuration, data and custom code come with the system, and the simplification items SAP publishes determine what must be changed.
Selective data transition (SDT). SDT covers approaches that build a new S/4HANA system but bring over a defined scope of data and configuration from the old one. SAP’s own material distinguishes two variants. In a shell conversion, you create a copy of the production system, strip its transactional data, convert the shell to S/4HANA and then load the data you want to keep. In mix and match, you build a new S/4HANA system and selectively transfer configuration and data from the legacy one. SAP offers this through its Data Management and Landscape Transformation services (DMLT), and more recently through a lean SDT approach built on the SAP Business Transformation Center.
Which paths each edition supports
| Target | New implementation | System conversion | Selective data transition |
|---|---|---|---|
| SAP Cloud ERP Public Edition | Yes (the only option) | No | No |
| SAP Cloud ERP Private | Yes | Yes | Yes |
| SAP S/4HANA on-premise | Yes | Yes | Yes |
Source: SAP Learning, “Describing the Customer Transition Paths”, and SAP’s transition paths guidance for user groups. A customer moving from ECC to Public Edition is therefore, by definition, doing a new implementation, even if it reuses some of its existing data and design decisions.
What differs between the editions
The path matters less than the target in several respects, so it is worth being explicit about what changes with the edition.
Release cadence. Public Edition delivers two major releases a year (for example 2602 and 2608) plus monthly updates (minor releases and hotfixes). All customers are on the same release and cannot defer the calendar, but many new functions in the continuous releases are switched off by default, so each customer can evaluate and activate them flexibly when ready. Private Edition and on-premise follow a two-year release cycle (base release), with non-disruptive minor updates through feature packages (FPS) every six months during the first two years, and then through support packages (SPS) every six months after the next base release ships, with standard mainstream maintenance for seven years. Although Private Edition and on-premise share the same technical development calendar, operational flexibility changes radically with the deployment model. On-premise, the customer has full control over its infrastructure. It can skip releases or stay on a specific one for years, provided it takes on the upgrade effort before the seven-year support period expires. In Private Edition, flexibility is agreed by contract: although the customer does not manage the infrastructure, it chooses the time window in which to apply updates, coordinating the process with SAP.
Extensibility. SAP classifies extensions in four clean core levels, A to D, and the editions differ in how many of them they make available (see our earlier series on clean core for the wider approach). A level is set by the lowest-ranked technology an extension uses.
| Level | Public Edition | Private Edition | On-premise |
|---|---|---|---|
| A: released APIs, ABAP Cloud, key-user, side-by-side | Yes, the only level | Yes | Yes |
| B: classic APIs, BAdIs, user exits | No | Technically yes | Technically yes |
| C: internal objects | No | Technically yes | Technically yes |
| D: modifications, implicit enhancements, noAPI objects | No | Technically yes, not recommended | Technically yes, not recommended |
Public Edition is limited to Level A: SAP’s own wording is that the public edition cannot be customized in its source code, only extended. Private Edition and on-premise run the same S/4HANA core, and in SAP’s learning material for both, classic ABAP extensibility remains available.
The recommendation is the same in every edition, and it is not to change standard code. Build at Level A by default. Where no released API exists, use classic enhancement points that SAP treats as stable (BAdIs, user exits), which is Level B. Treat Level C as an exception that needs governance and the changelog check. Level D is a remediation backlog, not an option for new work. A system conversion brings existing custom code and modifications along, so much of it will sit at C or D; the clean core work after a conversion is largely moving it up. New implementations start at A and can stay there.
Operations and responsibility. In Public Edition, SAP runs the system, upgrades it on its schedule and you configure within the scope of the offering. In Private Edition, SAP (or its hosting partner) runs infrastructure and basis under the RISE contract, while you retain control over upgrade timing within the release cycle and over custom code. On-premise leaves infrastructure, basis, upgrades and security with you.
Innovation and AI: what each target allows
AI is now part of the edition decision, and SAP’s position has shifted since 2023, so we separate what SAP documents from what the press reports.
Public Edition. Joule is activated through guided onboarding in which SAP automates most of the setup (the Joule Booster). The customer provides an SAP BTP global account with Joule and SAP Build Work Zone entitlements, SAP Cloud Identity Services, a BTP subaccount and administrator access. Tenant isolation and lifecycle are managed by SAP. New capabilities arrive with the half-yearly and monthly releases described above.
Private Edition. The BTP and identity prerequisites are the same, but more of the work sits with the customer: secure connectivity between BTP and S/4HANA (for example through Cloud Connector), identity integration, content exposure and role design. SAP Note 3523238 lists what Joule can do in Private Edition. According to CIO.com (May 2026), SAP will activate up to three Joule assistants at no additional cost for customers on its Private offering, covering common finance, HR and supply chain workflows.
On-premise S/4HANA and ECC. In July 2023 SAP’s CEO told investors that SAP’s newest innovations would be delivered only in the public or private cloud under RISE with SAP (The Register). At Sapphire in May 2026, SAP executives said most Joule assistants and agents would also work with ECC and on-premise S/4HANA, as an interim solution for customers that have committed the majority of their landscape to the cloud journey (The Register, CIO.com). The practical consequence is that choosing on-premise as the destination keeps SAP’s own AI offer tied to a cloud commitment.
Commercial model. SAP describes two tiers: Joule Base is included in standard SAP Cloud subscriptions, and Joule Premium is priced per user or through consumption of SAP AI Units (SAP). What is included in a specific RISE contract varies, and the advisory sites that publish breakdowns disagree with each other, so entitlements should be confirmed in writing before the business case counts on them.
| Target | How AI is delivered | Customer effort | Condition |
|---|---|---|---|
| Public Edition | Joule, activated with SAP’s guided onboarding; continuous releases | Lowest: BTP account, identity, subaccount | Joule Base included; Premium by user or AI Units |
| Private Edition | Joule; up to three assistants at no extra cost (as reported) | Higher: connectivity, identity, role design | Depends on RISE contract |
| On-premise / ECC | Joule assistants and agents in a hybrid setup (as reported) | Highest: BTP setup plus hybrid connectivity | Cloud commitment; terms reported differently |
Why the path matters for AI
Assistants and agents act on the system’s processes with the user’s existing SAP authorizations (SAP states that Joule upholds role-based access controls). That makes three things relevant to path choice: how standard the processes are, how clean the data is, and how well roles are designed.
On the first, an IDC analyst quoted by CIO.com (September 2026) says agentic AI needs “a single, trustworthy version of the process to act on” and that customization fragments it. The same article cites an IBM study (November 2025) that found unaddressed technical debt can consume 18% to 29% of AI implementation costs. Treat both as analyst views, not measured outcomes of S/4HANA projects.
The paths differ accordingly. A new implementation built standard-first with extensions on released APIs gives the cleanest base for AI, at the cost of the redesign effort described earlier. A conversion to Private Edition can use Joule as documented, but the custom code, modifications and data quality come with it, and the value you get depends on how much you remediate. SDT lets you leave behind selected code and data without a full reset.
Technical differences by path
| Aspect | New implementation | System conversion | Selective data transition |
|---|---|---|---|
| Starting point | New, empty system | Existing production system | New system, or converted shell of production |
| Data carried over | Master data, open items, balances (scope is chosen) | All data and history | Chosen scope, by company code, period or other selection |
| Configuration | Built new (fit-to-standard) | Existing, adapted to simplifications | Selectively transferred or rebuilt |
| Custom code | Rebuilt selectively | Analyzed and adapted (ATC, custom code migration) | Selectively migrated |
| Main tools | SAP Activate, Migration Cockpit | Maintenance Planner, SAP Readiness Check, simplification item check, SUM with DMO | SAP DMLT, SAP Business Transformation Center, SAP LT |
| Downtime | Cutover window, parallel build | One technical conversion run; dual-landscape retrofit if development continues | Cutover window; depends on variant |
| Main technical risk | Data completeness and cutover | Legacy complexity, custom code, data quality carried forward | Scope definition, tooling, data consistency |
Some detail on the other two paths (the conversion toolchain is covered in the next section):
New implementation. The Migration Cockpit loads data into the new system. It supports direct transfer from an SAP source system (since release 1909) and staging tables (since 2020). Fit-to-standard workshops replace the older blueprint phase: the team demonstrates standard processes and records the gaps, instead of documenting requirements from a blank sheet.
SDT. The scope selection is the technical core: which company codes, which time slice, which master data, which configuration. Shell conversion preserves configuration and custom code while removing data; mix and match rebuilds the system and transfers what is selected. Both depend on the consistency of the selected data across modules, and it is the main source of risk in SDT.
Technical prerequisites and existing code
This is where conversion and selective data transition projects differ most from a new implementation, so it deserves more detail.
Start release and technical baseline. SAP states the following for a single-step conversion to S/4HANA 2025: the system must be on Unicode (a single character encoding that covers all languages, as opposed to language-specific code pages); dual-stack systems, which run ABAP and Java in the same system, are not supported and need a dual-stack split first; and the start release must be SAP ERP 6.0 with enhancement packages 0 to 8, with restrictions for some support package levels listed in SAP Note 3549655. SAP Suite on HANA, Simple Finance 1.0 and 1503, and S/4HANA Finance 1605 are also supported start points. Earlier Business Suite releases need a two-step approach. In practice this means no mandatory upgrade to EHP8 before converting, but you must check your support package level against the restriction note. The technical conversion runs through SUM 2.0, which also migrates the database if the system is not yet on SAP HANA. Maintenance Planner calculates the target stack, and SAP Readiness Check covers add-on compatibility, sizing and a first custom code view.
Finance and the new General Ledger. You do not need to migrate to the new General Ledger in ECC before converting. A system on classic G/L can be converted directly, and the conversion moves classic ledger 00 into the leading ledger 0L of the Universal Journal. SAP refers to Note 2261242 for the finance conversion paths. Document splitting and additional ledgers cannot be introduced as part of the conversion itself. A company that needs them must either run a separate new G/L migration in ECC first, or introduce them after the conversion. That is a business decision about finance design, not a technical prerequisite.
Business Partner and CVI. This one is mandatory and happens in ECC, before the technical conversion. In S/4HANA the customer and vendor masters are replaced by the Business Partner as the lead object, and the Customer/Vendor Integration (CVI) keeps the two in sync. Before the conversion every customer and vendor must have a corresponding Business Partner. In practice this means: activate the CVI configuration and business functions in ECC, clean the customer and vendor data, synchronize it with the CVI/MDS load cockpit, and verify with the completeness check (transaction CVI_COMPL_CHK). Where a customer and a vendor are the same legal entity, you decide whether to merge them into one Business Partner; that requires a custom BAdI implementation (CVI_CUSTOM_MAPPER), because the standard mapping only matches on cross-references and identical address data. The legal entity harmonization report CVI_LEDH is available only from ECC 6.0 EHP8 SP17. It is a convenience, not a requirement, but it matters if you plan to merge customers and vendors.
Existing custom code. SAP splits the work into four steps.
- Scoping. Identify which custom code is actually used and delete the rest through deletion transports during the conversion. This shrinks the adaptation effort and is worth doing first.
- Analysis. Run the ABAP Test Cockpit (ATC) with the check variant S4HANA_READINESS_REMOTE from a central check system (SAP_BASIS 7.52 or higher, or an S/4HANA system from 1809), using the Simplification Database imported via transaction SYCM (SAP Note 2241080). The result is a list of findings that reference the simplification items and the SAP Notes that explain each fix.
- Adaptation. SAP places the functional adaptation after the technical conversion. Modifications to SAP objects are handled in SPDD and SPAU during the conversion; the remaining findings are fixed afterwards in ADT (ABAP Development Tools for Eclipse), partly through semi-automated quick fixes.
- Performance tuning. Once on HANA, the most expensive SQL statements are reviewed and optimized.
The key consequence for the path decision: in a conversion, all custom code and all modifications come along and must be assessed. Code that reads or writes tables whose data model changed (for example finance tables replaced by the Universal Journal) will show up in the ATC results. Whether you do much of the adaptation before the conversion (scoping, deletion, early ATC runs) or after it is a project planning choice, and SAP recommends running Maintenance Planner, the simplification item check and the custom code worklists early. If the business keeps changing the old system during the project, a dual landscape with retrofit is needed, which adds effort and its own error risk.
How this changes by path.
| Topic | Conversion | New implementation | Selective data transition |
|---|---|---|---|
| Start release and Unicode | Must meet the rules above | Not applicable to the target; the source only needs to be extractable | Shell conversion: same rules as conversion. Mix and match: not applicable to the target |
| CVI and Business Partner in ECC | Mandatory before conversion | Not needed in the source; Business Partners are created in the target by the migration | Shell conversion: mandatory. Mix and match: not needed in the source |
| Custom code | All code is carried over and must be scoped, analyzed and adapted | Rebuilt selectively, ideally on released APIs and ABAP Cloud | Shell conversion: carried over and adapted. Mix and match: selected and adapted |
| New G/L | Not required beforehand; converted automatically | Not applicable; configured new | Depends on variant |
How the project approach differs
The labels also describe different kinds of projects, and the difference shows up in organization more than in technology.
Conversion is largely a technical project with a business validation phase. The scope is defined by what the existing system does. Business involvement concentrates on testing. The timeline is generally the shortest of the three, but the result is the legacy process landscape on a new platform, and the clean-up of custom code and data quality still has to happen, before or after go-live. Projects that skip it carry their technical debt into the new system.
New implementation is a business project that uses technology. The process design is the main effort. Fit-to-standard forces decisions on which standard to adopt and which gaps justify an extension. It takes longer to start delivering value, because the organization must agree on its target processes, but it is the only path that lets you reset data and custom code at the same time. Data migration is a project in its own right and needs business owners for data quality.
SDT is a project defined by its scope. Someone has to decide what is kept, and that decision needs governance. It is attractive when parts of the business must remain stable (a legal entity, a history requirement) while others are restructured, for example following an acquisition or divestiture. The cost is design effort in selection logic, plus dependence on specific tooling and services.
Across all three, SAP Activate provides the methodology (discover, prepare, explore, realize, deploy, run), with SAP Cloud ALM for project and operations management. What changes between paths is how much of the explore phase is a design exercise versus a technical assessment.
The deadline
SAP’s support portal states that mainstream maintenance for SAP Business Suite 7 (including ECC 6.0 with enhancement packages) ends at the end of 2027. Extended maintenance runs from 2028 to 2030 and costs two percentage points more in maintenance fees. After 2030, only customer-specific maintenance is available.
On whether SAP will move the date: CIO.com reported in May 2026 that an SAP spokesperson said the company has no plans to extend the ECC end-of-service deadline beyond 2027, with extended support through 2030 at additional cost.
The timeline matters for path selection. For durations we have one sourced figure: an IDC survey from May 2025, cited by CIO.com, puts SAP HANA migrations alone at roughly 15 months on average, before any custom code work. The rest is our judgment from practice: a conversion is generally the shortest path, and a new implementation with process redesign the longest. A customer starting now, in late 2026, and aiming to be live before the end of 2027 has little room for anything beyond a conversion or a tightly scoped SDT, and otherwise should budget for extended maintenance.
Recommended path by case
These are our recommendations, based on SAP’s guidance and the facts above. They are judgments, not rules; the right answer depends on your landscape.
| Situation | Recommended path | Reasoning |
|---|---|---|
| Want Public Edition; standard processes acceptable | New implementation (only option) | Conversion and SDT are not offered for Public Edition |
| Stable ECC, low customization, short deadline, no process change planned | Conversion to Private or on-premise | Fastest route; low risk of carrying much debt. Prefer Private if AI is likely later (see below) |
| Heavy custom code, high modification rate, clean core is a goal | New implementation, or SDT with selective custom code migration | Conversion carries the debt; the cost of cleaning later is higher |
| Poor master data quality, many legacy duplicates | New implementation | A clean data load is the most direct way to leave the problem behind; SDT can do it only for the scope you choose not to carry over |
| Multiple ECC systems to consolidate into one S/4HANA | New implementation or SDT (mix and match) | Conversion keeps each system separate; consolidation needs a new target |
| Acquisition or divestiture in progress; one entity must be carved out | SDT | Scope selection by company code is what SDT is designed for |
| Long transaction history required online for legal or audit reasons, and no appetite for archiving | Conversion or shell-conversion SDT | These preserve history in the system; a new implementation does not |
| Mixed landscape: a stable core plus subsidiaries that can run standard | Two-tier: Private or on-premise at headquarters (conversion), Public for subsidiaries (new implementation) | Matches the path to each entity’s needs |
Most real cases combine several of these situations, so the path should follow the strongest constraint. Conversion is the right choice when the existing system is in reasonable shape and the business is not asking for change. A new implementation is only worthwhile if the organization redesigns its processes; one that recreates the old processes delivers little more than a conversion, at higher cost. For innovation and AI, see the next section.
If innovation is the priority, and if it is not
Whether AI and innovation are the main goal of the project changes how the same three paths are used. These are our recommendations, not SAP’s.
When innovation is a main goal.
- Standard acceptable, no heavy localization: new implementation on Public Edition. It has the shortest route to continuous AI delivery and the lightest Joule setup.
- Standard not enough, or regulated landscape: Private Edition with a new implementation or SDT, ABAP Cloud for new development and side-by-side extensions on BTP, so the core stays clean.
- Deadline forces a conversion: convert to Private Edition, then run a clean core remediation as a planned second wave, with a target for how much custom code reaches the better clean core levels. Do not treat AI activation as the finish line of the conversion.
- In all cases, agree three to five AI use cases in the explore phase rather than after go-live. Finance close and reconciliation, procurement contract management and supply chain demand planning are the areas where, according to IDC as quoted by CIO.com, agentic AI shows payoff first. Activate Joule in a non-production system during the realize phase to test roles and connectivity.
When innovation is not the priority now. These steps cost little and keep AI possible later:
- Choose the path on business and deadline criteria, as described above.
- Stop adding modifications. Apply clean core rules to all new development from the start of the project, so the system does not accumulate new debt that AI would later inherit.
- Do the data work the migration already requires (Business Partner and CVI, duplicate clean-up, archiving). It is needed for the conversion and is also the data foundation AI will use later.
- Put identity and BTP basics in place early: SAP Cloud Identity Services and a BTP global account are prerequisites for Joule on both cloud editions, and are small tasks compared with the migration itself.
- Favor an edition that leaves AI possible later. If AI is plausible in the next two to three years, Private Edition keeps SAP’s own AI offer available without a further cloud commitment; on-premise ties it to one.
- Check the AI entitlements in the RISE contract and note what would be billed in AI Units.
- Plan an innovation wave after stabilization, with an owner, a use-case backlog and measures, instead of leaving AI as an open-ended future item.
Our view
Choose the target first, then the path. If Public Edition is acceptable to the business, there is no path decision: it is a new implementation. If you need Private or on-premise, the choice is between keeping what you have (conversion), starting clean (new implementation) or choosing what to keep (SDT).
Conversion fits when the system is sound and the deadline is tight. New implementation fits when the business wants to change and the data or code is a liability. SDT fits when both apply, provided you have the governance to define the scope and accept that it is the least standardized of the three.
If innovation is a main goal, the decision is mostly about the target: a standard-first Public Edition or a clean-core Private Edition. If it is not, the path can follow business and deadline criteria, provided you apply clean core rules, finish the data work and prepare identity and BTP early. In any case, set the clean core target before the project starts: a conversion that leaves custom code unaddressed carries its technical debt into S/4HANA.
Sources
SAP Learning: Describing the Customer Transition Paths to SAP S/4HANA Cloud Private Edition
SAP Learning: Analyzing System Conversion
SAP Learning: Explaining the Extensibility Possibilities Depending on the SAP S/4HANA Version
SAP Learning: Exploring Clean Core Extensibility Best Practices (clean core levels A to D)
SAP News Center: Discover How to Extend SAP S/4HANA Cloud the Right Way
ASUG: SAP Updates Clean Core Guidance to Help Customers Clarify Extensibility Criteria
SAP: Discover Transition Paths to SAP S/4HANA (user groups)
SAP Support: Maintenance strategy for SAP Business Suite 7 and SAP S/4HANA
SAP Help Portal: Getting Started With SAP S/4HANA 2025
SAP Learning: Performing the Custom Code Analysis
ERPfixers: SAP ECC Classic GL to SAP S/4HANA direct conversion (vendor source)
SAP News: Serving Customers Better with Lean Selective Data Transition
SAP News: Embracing the Strategic Shift to Cloud ERP
SAP News: SAP S/4HANA Migration Guide for IT Executives
SAP News: SAP Cloud ERP Private FPS01
SAP Blogs: Move to SAP S/4HANA with Selective Data Transition
SAP Community: Key Differences Between SAP S/4HANA and SAP Cloud ERP Private
ASUG: How IBM’s Bill Piotrowski Sees Selective Data Transition Driving SAP Success
TIMETOACT: SAP Simplifies the Naming of its Cloud ERP Offering (partner source)
SAP Learning: Setting Up Joule for SAP S/4HANA Cloud Private Edition
SAP Learning: Setting Up Joule for SAP S/4HANA Cloud Public Edition
SAP: What is Joule? (product page, including pricing tiers)
The Register: SAP U-turn brings AI features to ECC and on-prem S/4HANA (press)
CIO.com: SAP’s AI offer to legacy customers comes with a catch (press)
CIO.com: The ERP reckoning: Decades of customization could block AI value (press)
