Executive Clarity · Architecture & Landscape

SAP instance strategy

One global core, regional cores, or a core per division — the landscape decision boards inherit for a decade. It is a business-model decision wearing a technical costume.

The question

Should the enterprise run one SAP instance for everything, one per region, or one per division — and by which criteria should the board decide, before anyone argues about deployment models?

The three patterns

  • Single global instance — the highest degree of process harmonisation: one system, identical procedures and configuration worldwide. Typically driven by globally integrated supply chains; common in chemicals and high tech.
  • Regional instances — a core per major region. Fits regional supply chains and reporting structures; common in consumer products. Each estate must still be kept in sync on processes, releases and master data.
  • Instances per division — a core per business. Fits conglomerates with strong process variation between divisions and few shared services across them.

Decision factors

The documented facts

  • SAP's enterprise-architecture guidance treats instance strategy as an explicit design decision with distinct harmonisation, cost and governance profiles per pattern — not as an implementation detail.
  • A single instance concentrates consistency, maintainability and reporting advantages — one chart of accounts, one controlling area, one process truth.
  • Multiple instances multiply integration, synchronisation and release-management effort: the estates must stay aligned on processes, patch levels and data integrity to remain comparable.
  • Data-residency and country-specific requirements can force local processing regardless of the preferred pattern — regulation constrains the design space before preference does.

Source caveat: pattern trade-offs are drawn from SAP's published architecture guidance and practitioner discussion. They describe tendencies, not guarantees — the decisive inputs are your operating model and your regulatory footprint.

Inference

  • The instance question is the operating-model question in disguise. Where processes are genuinely shared, one core makes them cheap; where they genuinely differ, one core makes them political.
  • Consolidation buys reporting truth and lower long-run cost at the price of slower local change. Fragmentation buys local speed at the price of permanent integration overhead. Neither is free.
  • M&A ambition belongs in the criteria: a company that buys and sells businesses needs a landscape that can absorb and release them — which argues for clean boundaries, whatever the pattern.
  • Decide the instance pattern before the deployment model. Public cloud, private cloud and on-premise are answers to a later question; an instance strategy derived from a licensing preference is backwards.

Point of view

Most instance debates are settled by history, not strategy — the landscape is whatever the mergers left behind. The honest starting point is to write down the six criteria below, score the three patterns against them in one workshop with business owners present, and record which criterion actually decided it. A landscape decision no one can reconstruct is a landscape decision that will be relitigated every budget cycle.

The six criteria

1

Fit to enterprise strategy

Alignment with business goals and digital ambition. Does the landscape support how the company intends to grow, acquire and comply?

2

Target operating model

Which processes are shared, which are local, and who decides? The instance map should mirror the answer, not fight it.

3

Risk to business & IT operations

Security, compliance and resilience. Concentration risk of one core versus the coordination risk of many.

4

Scale & performance

Sizing for current volume and future growth, load distribution, and the data-residency constraints that bind the design.

5

Change management

Governance of transformation and operations: release trains, template ownership, and the training burden each pattern creates.

6

Long-term cost

Lifecycle TCO — infrastructure, maintenance, licensing and operations — plus the investment protection of an architecture that survives reorganisation.

Monday-morning questions

  • Which processes are genuinely global today — measured by shared configuration, not by org-chart aspiration?
  • If we divested a division next year, how long would it take to carve its data and processes out of the landscape?
  • Which countries in our footprint impose data-residency or localisation requirements that constrain the pattern?
  • Who owns the global template, and what is the documented process for a region to deviate from it?
  • What does each pattern cost over seven years — including the integration and synchronisation work the slideware omits?

Where this sits in the decision chain

Instance strategy precedes the Public vs Private Cloud decision and frames the two-tier ERP pattern — the hybrid answer for headquarters-plus-subsidiaries structures. The governance it demands is the subject of transformation governance; its economics belong in the business case & TCO decision.

Sources

Editorial standard: vendor guidance, inference and point of view are kept separate above. Published 2026-08-03 · By Andreas BORN, whose SAP career included system-landscape-optimization leadership across mergers, divestitures and consolidations.