Introduction

Most discussions about SAP BW migration in 2026 are happening in the wrong order. The CFO is being asked to choose between BW/4HANA, SAP Datasphere, and Snowflake. The CIO is producing comparison decks. Procurement is running RFI processes against three vendors. Everyone is doing the work of selection — and almost no one is doing the work of inventory.

The result is a particular kind of corporate planning failure. The decision gets made on incomplete information, the programme is scoped against an estate nobody has actually mapped, and somewhere in month nine the programme team discovers that the custom ABAP volume is forty per cent higher than the estimate. The fixed-price proposal becomes a change request. The 2027 deadline starts to feel uncomfortably close. By then, the only thing the CFO can do is approve the cost overrun.

This article is for finance and programme leaders who have not yet started — or who started but have not yet found the planning data their decision needs to rest on. It makes three points.

The arithmetic

SAP BW mainstream maintenance ends in 2027. That is a fixed point in time. What is not fixed is how long a migration takes — and the planning assumption that should sit at the top of every CFO's mental model is that a safe migration of a large BW estate takes eighteen months.

That number is not generous. It assumes a structured programme with phased rollout, business-by-business reconciliation, parallel running before cutover, and a hypercare period long enough for the internal team to absorb the new platform. Compressed timelines exist — but they trade safety for speed, and the trade is almost always made under duress because the planning started late, not because compression was the right design choice.

Most organisations who believe they have "about a year" to make a decision actually have zero runway. The decision is the entry into the planning, not the planning itself.

The arithmetic that follows is uncomfortable. If a large migration takes eighteen months, an organisation aiming to go live before the 2027 maintenance cutoff needs to start the build no later than mid-2026. To start the build, the organisation needs an approved business case, a scoped programme, and a contracted delivery partner. To produce the business case, the organisation needs an inventory of what it owns and a credible estimate of what migration will cost — both of which require time.

Three paths — and why the comparison is less even than it looks

The standard frame for SAP BW migration is three paths: stay in the SAP ecosystem with BW/4HANA, move to SAP Datasphere, or migrate to Snowflake. Vendor sales motions encourage the framing as a balanced comparison — three roughly equivalent options each with their trade-offs. The honest TCO comparison is less balanced.

BW/4HANA retains the SAP licence dependency. It retains the HANA infrastructure cost. It retains the requirement for ABAP talent at the same time as the supply is contracting. Five-year TCO is the highest of the three options. The argument for it is continuity — but continuity with what? The platform reaches its own end-of-maintenance horizon in 2032. Organisations choosing BW/4HANA in 2026 are setting up to have this same conversation five years later, with the same talent constraint and the same migration cost, only with their decision-making muscle weaker because they deferred it once already.

SAP Datasphere is the modernisation story SAP has been telling since 2023. It improves on BW/4HANA in important ways, but it retains the SAP licence dependency and adds BTP costs over time. ABAP is still partially required. For organisations with deep SAP investment beyond analytics — strong S/4HANA, significant SAP application footprint — there is a defensible case. For organisations whose SAP dependency is primarily BW, Datasphere is moving sideways at substantial cost.

Snowflake severs the dependency. The five-year TCO reduction versus BW/4HANA can fall in the range of forty to sixty per cent. The ABAP skills requirement is eliminated post-migration. The talent pool shifts from a contracting set of ABAP specialists to the substantially larger pool of SQL engineers. AI and real-time analytics — the capabilities that increasingly drive commercial differentiation — are native to Snowflake in a way they are not, and likely will not be, on the SAP analytics stack.

The comparison is not even on a five-year view. The reason it features even in most conversations is that the comparison is being presented in the wrong dimensions. Migration complexity is real and matters. Talent dependency, licence cost, and platform longevity matter more — and they all point in the same direction.

The compounding cost most CFOs are not modelling

The talent question deserves separate treatment because it is the one variable that gets materially worse the longer the decision waits.

Experienced SAP BW and ABAP practitioners are leaving the market. New engineers are not being trained on a platform approaching end-of-life. The day rate for an ABAP specialist is already running at approximately two thousand US dollars and rising as supply contracts. For organisations with significant custom ABAP transformation logic — which is most BW estates — this creates a compounding cost problem.

The implication for CFOs is that the cost model used to evaluate "migrate now" versus "migrate later" needs to include a forward curve on ABAP cost, not just a present-day estimate. An organisation that defers the migration by twelve months is not buying twelve months at today's cost; it is buying twelve months at next year's cost, with fewer specialists in the market and a tighter delivery timeline forcing a price premium.

This is also the strongest argument against the BW/4HANA path. The path that retains the ABAP dependency is the path most exposed to the cost curve that is moving in the wrong direction.

The missing piece — and what changes that

If the arithmetic is pressing and the path is clearer than most boardrooms acknowledge, why are not more organisations making the decision? In most cases, because the planning information needed to make the decision credibly does not yet exist.

Most BW estates have grown organically over a decade or more. Active InfoProviders sit alongside dormant ones that nobody dares delete. Custom ABAP transformation logic has accumulated, written by people who often no longer work at the organisation, documented inconsistently if at all. BEx queries and reports run in production whose business owners are unclear. Without an inventory of what is actually in the estate, programme scoping is guesswork — and CFOs are right to be cautious about approving large fixed-price commitments based on guesswork.

This is the gap the Forge two-week landscape assessment is built to close. It connects to the existing BW system via RFC and produces a complete inventory: every InfoProvider classified as active, dormant, or obsolete based on actual usage metadata; every custom ABAP routine catalogued with complexity scoring; business domain dependencies mapped; a phased migration roadmap sequenced by complexity and priority; a five-year TCO comparison across all three paths built on the actual landscape.

The assessment is priced as a fixed-fee engagement designed to sit within single-approver budget thresholds. The deliverables are useful regardless of which path or partner the organisation ultimately chooses — they answer the inventory question, which is the precondition for any credible migration decision.

Next step

The right next step for most CFOs is not to commission another vendor evaluation. It is to commission a credible inventory of the estate they are about to make a decision about. The five-minute readiness check at jarvisbusiness.io/bw-assessment.html is the entry point: it confirms BW is in scope and indicates the engagement tier. The two-week assessment that follows produces the data the boardroom decision needs to rest on.

The 2027 deadline is fixed. The clock is running. The cost of waiting is not just the cost of a later migration — it is the cost of a worse one, made under tighter timeline pressure, with fewer specialists available to deliver it, at higher day rates.

There is still time to plan this properly. There is not enough time to plan it twice.