Case Study 2: Commonwealth Bank of Australia's Core Banking Overhaul

Background

In the mid-2000s, the Commonwealth Bank of Australia (CBA) — one of the country's "Big Four" banks — made a bold decision: it would replace its core banking system. The existing system, like those of most major banks, was a COBOL-based mainframe application that had been in production for decades. CBA chose to replace it with SAP's Banking Services on HANA platform, making it one of the first major banks to attempt a full core banking replacement.

The Project

The project, known internally as the "Core Banking Modernisation" initiative, began in 2008. The original timeline was approximately four years. The original budget was approximately $580 million AUD.

The actual timeline extended to approximately five years, with the final migration of customer accounts completing in 2012. The actual cost was approximately **$750 million AUD** (roughly $1 billion USD at the time), though some estimates including indirect costs and organizational change management push the total past $1.5 billion AUD.

Key statistics: - 9 million customer accounts migrated - 5 years from initiation to completion - $750 million AUD in direct costs (budget overrun of approximately 30%) - Thousands of COBOL programs replaced - Zero downtime during the final migration weekend (a point of legitimate pride)

The Business Case

CBA's rationale for replacement rather than modernization included:

  1. Product agility: The old system made it difficult to launch new banking products quickly. Each new product required extensive COBOL programming and testing.
  2. Real-time processing: The batch-oriented COBOL system processed transactions in cycles. CBA wanted real-time transaction processing for competitive advantage.
  3. Customer experience: Modern digital banking interfaces were constrained by the capabilities of the back-end system.
  4. Talent pipeline: CBA anticipated the same COBOL talent gap that this chapter discusses and wanted to reduce its dependency on scarce mainframe skills.

The Results

CBA's migration is widely regarded as a success — with significant caveats:

Successes: - CBA achieved faster product launches (new products in weeks rather than months) - Real-time payment processing gave CBA a competitive advantage in the Australian market - The bank's digital banking platforms improved significantly - CBA reduced its mainframe footprint and associated licensing costs

Challenges: - The 30% cost overrun was substantial - The project required years of parallel running (old and new systems simultaneously), which doubled operational complexity - Some complex business logic from the COBOL system was difficult to replicate and required significant rework - Organizational change management was as challenging as the technical migration

Analysis Questions

  1. CBA spent approximately $750 million to replace its COBOL core banking system. Using the salary data from Section 1.4, calculate how many years of COBOL developer salaries that amount could fund. What does this comparison tell you about the economics of maintain-vs-replace decisions?

  2. CBA cited "talent pipeline" as one reason for migration. Is reducing dependency on COBOL skills a valid business reason for replacement, even if the existing system works well? Under what circumstances does this argument become compelling?

  3. Most banks that have studied CBA's example have not followed suit. Why might other banks reach a different conclusion about replacement even after seeing CBA's success?

  4. The project took five years and cost 30% more than budgeted. These numbers are actually better than average for large-scale core system replacements. What does this tell you about the feasibility of replacing COBOL systems across the industry?

  5. How does CBA's story illustrate the Modernization Spectrum discussed in Section 1.9? Where on that spectrum did CBA fall, and what were the consequences of that choice?

Connection to Running Examples

GlobalBank, our running example, has discussed a CBA-style replacement multiple times over the past decade. Each time, the analysis has concluded that the cost and risk are too high. Instead, GlobalBank has pursued an "encapsulate and extend" strategy — keeping the COBOL core but wrapping it in modern APIs and building new capabilities in hybrid architectures. Priya Kapoor, the application architect, led this strategy, and it serves as the backdrop for many of the integration examples in Part VIII of this textbook.

MedClaim, by contrast, has considered a more aggressive modernization path for its claims adjudication engine but has been constrained by regulatory requirements that make system changes in healthcare especially risky and slow.

Lessons for COBOL Professionals

  • Replacement is possible but extraordinarily expensive. CBA's $750 million price tag — for a single bank in a single country — illustrates why most organizations choose maintenance over replacement.
  • Success is measured in years, not months. Even a well-funded, well-executed replacement takes years. During those years, both the old and new systems must be maintained.
  • Business logic is the hard part. The COBOL code itself is not the primary challenge in migration — it is the business rules encoded in that code, accumulated over decades, that are difficult to replicate.
  • The decision is economic, not technical. CBA did not replace COBOL because COBOL is bad. It replaced COBOL because the business benefits (product agility, real-time processing, talent strategy) justified the enormous cost. For most organizations, the math does not work out the same way.