Lesson 3 of 5
Analytics placement
9 min
Listen to this lesson
Every SAP program argues about where reports should run. Put everything in S/4HANA and you risk overloading the transactional system and missing non-SAP data. Put everything in a warehouse and operational users wait for loads and reconcile two versions of the truth. The architect's job is to place each analytical need where it fits best, and to make that placement deliberate.
Embedded analytics in S/4HANA
SAP describes embedded analytics in S/4HANA in three layers: the SAP HANA database holding the business data, a virtual data model built from Core Data Services (CDS) views, and SAP Fiori analytical apps on top. Analytical queries built on CDS views feed multidimensional reports, KPIs and SAP Analytics Cloud stories through a live connection. Because the data is read where it is created, embedded analytics is real-time and uses the system's own authorizations.
The data platform
A separate analytics platform earns its place when data comes from many systems, when history must be kept beyond what the operational system holds, when data must be harmonized across sources, or when heavy analysis would compete with transactions. SAP's offering here centers on SAP Business Data Cloud, which brings together SAP Datasphere (described by SAP as a business data fabric for integrating, modeling and distributing data) and SAP Analytics Cloud, with the concept of data products: packaged, governed, reusable data sets owned by a domain team.
Data moves from S/4HANA to the platform through extraction-enabled CDS views and replication. SAP Datasphere's documentation lists, per SAP S/4HANA release including 2025, the SAP Notes that must be applied before replication flows can use an S/4HANA system as a source. That is a typical architecture dependency: the analytics design creates technical prerequisites in the ERP system that Basis must plan.
| Need | Embedded in S/4HANA | Data platform |
|---|---|---|
| Open sales orders at Detroit, right now | Best fit: real time, same authorizations | Lags behind the source |
| Five-year warranty trends across ERP, MES and portal | Limited: only S/4HANA data | Best fit: history and multiple sources |
| Daily credit exposure for Great Lakes Fleet Services | Best fit: operational, current | Possible but adds latency |
| Group margin including Alternadores before and after acquisition | Partial: only post-load data in company code 1910 | Best fit: harmonizes legacy history with S/4HANA |
| Ad hoc data science on large volumes | Risk of load on the transactional system | Best fit |
A placement rule set
- Operational, real-time and single-system → embedded analytics in S/4HANA.
- Cross-system, historical, harmonized or heavy → the data platform.
- Planning that writes back plans → the analytics platform's planning functions, with clear ownership of plan data.
- Never build a second copy of an operational report in the warehouse "just in case"; two versions of the same KPI will disagree.
- For each KPI, name one authoritative definition and one place it is calculated.
What would you do?
Dan wants a daily dashboard of open AR by customer for company codes 1710, 1810 and 1910, plus a three-year trend that includes Alternadores' legacy history. Where should each part live?
- Embedded analytics: HANA, CDS virtual data model, analytical queries and Fiori apps; real time and same authorizations.
- A data platform fits cross-system, historical, harmonized or heavy analytics.
- Replication from S/4HANA creates prerequisites in the source system.
- One definition and one calculation place per KPI; reconcile to the source.
Check your understanding · not graded
Lesson check
1. Embedded analytics in S/4HANA is best for:
2. What forms the virtual data model in S/4HANA embedded analytics?
3. Which need most clearly belongs in a separate data platform?
4. Why name one calculation place per KPI?