Don't Crack the Bank

Don't Crack the Bank

Core chaos

The core banking system (“CBS”) is both the brain and heart of any bank. It’s the system of record for most banking products, including customer information, savings, deposits, transaction, and loans. Crack it, then, you risk breaking your bank. And your executives will be thrown to the exit.

Several British banks found chaos trying to change their core systems. Having grown up in pre-Brexit England, this isn’t a peculiarly British problem. Taking a traditional approach to core banking system replacement in the U.S. makes the replacement decision dangerous.

Cobol cowboys save millenials

Almost all banks (including the U.S.) need to replace their core banking systems for five reasons.

First, they cost too much. Non-bank architectures have much lower cost-to-income ratios. Banks spend, on average, 11% of their revenue on IT. This exceeds most industries. Despite all the digital transformation declarations, only 30% of these IT budgets are spent on new things. The rest is running the place.

Second, current core banking systems don’t deliver the experience demanded by digital customers.

Some core systems are older than their millennial customers. 30-year-old-plus systems are common.

These ancient architectures were designed around products, not the customer. Originally designed for batch processing, these systems heave and creak to provide real-time information and a unified customer view.

Also, the typical core banking system hampers the employee experience too. These systems starve employees of the analytics and intelligence they need to serve their customers well. Duplicated data across product-siloed systems compound the challenge.

43% of American bank’s core systems are written in the 60-year-old development language, COBOL.

Third, longstanding implementations of core banking systems are technologically obsolete and anti-agile. 43% of American bank’s core systems are written in the 60-year-old development language, COBOL. Changing these systems is slow, painstaking work. Real “Cobol cowboys” are in high demand and don’t need loft-style offices. The typical bank struggles to shuffle out a new release cycle quarterly or half-yearly. In contrast, Google and Facebook make new releases every week.

Finally, retaining old systems is an unsafe choice. Sticking with the status quo raises the risks of acute, catastrophic IT problems. The U.K.’s Royal Bank of Scotland (RBS) is infamous for this — along with former CEO, “Fred The Shred”. I will discuss how this bank cracked later in this post.

Why try?

Big prizes await any bank that can replace its core banking systems successfully.

On costs, a new core banking system will boost a bank’s efficiency ratio. Morgan Stanley believes, “migrating to a state-of-the-art core banking system could reduce cost/income ratios by 9%.” Straight-through processing, real-time information, and extensive task automation will drive the improvements.

A modern core banking system can enhance customer experience dramatically. First, it provides a single, channel-neutral user interface for core banking services. Second, it enables the creation of new products and services with deeper customer data. Third, it helps incumbent banks build muscles to fight their FinTech challengers and new competitors, such as the Amazon army or Apple’s live-your-life on an iPhone.

Also, a state-of-the-art core banking system helps incumbent banks make the most of their key asset — their distribution network and customer base. For example, Chase touches around the same number of households as the combined populations of England and France.

Finally, the data science work that precedes a core banking system replacement helps get the bank ready for even newer technologies, such as machine learning and blockchain (once the use cases sharpen up).

The billion dollar decision

However, even successful core banking system replacements take lots of treasure and time — and successful, public examples are rare.

The best known success is Commonwealth Bank of Australia, with 1,131 branches. Working with Accenture, migrating to a SAP core banking system cost close to USD 1 billion (AUD 1.3 billion) and took five years. In the U.S., a public success is BBVA Compass, with 672 branches. Back in 2013, this was heralded as the first core U.S. core system replacement in 10 years. Even so, migration to Accenture’s Alnova system cost $360 million. The migration took three years.

Ways to crack a British bank

The experience of British banks show the bank-cracking risks of changing core banking systems. British banks hurt themselves trying the three main ways to address the core banking system replacement challenge.

Way 1: The in-house system option. TSB.

This core system change “put the bank on its knees,” according to TSB CEO, Paul Pester.

Customers were locked out of their accounts and other customers enjoyed access to strangers bank accounts. TSB said the IT problems cost it GBP 176.4 million in customer compensation, foregone income, waived fees, and fixing the system’s defects. 

The original plan was to shift the accounts of its 5.4 million customers from an older IT system. After separating from Lloyds Banking Group in 2013, TSB was sold to Sabadell, a Spanish bank. Post-separation, TSB continued to use and rent the Lloyd’s banking platform at an annual cost of USD 130 million (GBP 100 million). Apart from the expense, TSB viewed the Lloyds system as old and complex. The decision was taken to migrate to Sabadell’s Proteo system, which had been used for multiple acquisitions. Next, this happened:

A version of Proteo customized for the U.K. and TSB, called Proteo4UK, was built using 800 software engineers and another 200 people helping with the migration.

Way 2: The third-party system option. The Co-Operative Bank.

The Co-operative Bank is a full-service retail bank, serving small and medium-sized enterprises.  In 2013, the bank had around 300 branches. Mixing system problems with a merger, the bank has lost a cumulative total of GBP 2.6 billion over five years. This is a “challenged” bank with a cost-to-income ratio of 103% in 2017.

Anyway, the purpose here is to look at their core banking system decision and deployment. The attempt to replace its legacy systems with Infosys’ Finacle ended up with the writing off of GBP 349 million.

Independent review by Sir Christopher Kelly concluded: “it is difficult to avoid concluding that in many important respects the programme was set up and implemented in a way which was almost bound to lead to failure.”

He went on to say: “It is critical for an organisation to understand the extent to which it is capable of managing large-scale change programmes. It is important to be realistic about the scale of projects undertaken,and the burden this places on the organisation.”

Long after the Lean Startup approach (2011) was popularized, the Co-Op Bank kicked off the project by gathering 18,000 business and system requirements.

Sir Christopher’s conclusion is damning:

“At the time the bank decided on this option [replatforming], no UK full-service bank had successfully replaced its core banking systems…On the face of it, the bank was an unlikely trail-blazer.”

Way 3: Keep the legacy system. The Royal Bank of Scotland.

In June 2012, RBS upgraded its core banking system software. This “IT Incident” affected at least 6.5 million customers, who were unable to make payments, drawdown loans, or make international payments.

Before a discount for early settlement, the U.K. regulator, Financial Conduct Authority, imposed a USD 80 million (GBP 60 million) fine. RBS also paid out GBP 175 million in customer compensation.

Rational replacement therapy

Traditional principles, practices, and vendors have turned core bank system replacement into a crack-the-bank risk and a career-ending opportunity for executives.

Thinking about replacing core systems rationally requires five elements.

First, is your core system a source of competitive advantage? It depends on your business strategy. If maximum efficiency is the goal, then, maybe not. If you want to differentiate with unique features and control the complete customer experience, then, yes, it does matter.

Second, management teams have to ask themselves brutal questions:

  • Can our team handle a project this complicated? Do we have a history of success in anything of comparable complexity and size?

  • Do we have system integrator partners with actual, practical experience? Will they share the risk and pain when things go sour? Talking transformation is easy, getting digital done really hard. See The Digital Transformation Maze.

  • How can we radically simplify the design and deployment? Not 18,000 requirements. Better to focus on one use case and figure out the five foremost features that drive adoption.

Third, face failure. Do away with wishful thinking and hopeful strategies. The history of core banking system replacement means you have poor odds for success. You need to plan for when things go wrong, so, they will go right.

Manage the project in an honest, direct way. Only hire principals who own their work, revere their craft, and pursue the highest standards.

Ask fundamental questions over and over again:

  • Strategy: are we doing the right thing?

  • Architecture: are doing things the right way?

  • Delivery: are we doing things well?

  • Value: are we getting the expected benefits?

Fourth, go deep and narrow. Start with small steps in a narrow scope. One line of business or a new digital banking initiative. Build the minimum viable solution. This is more than the minimum viable product. It covers all the people, process, and change challenges as well. And a cloud-based solution is the only way to get things done at a tolerable speed.

Finally, consider new approaches, technologies, and providers, which I discuss below.

Incumbents relax

Leaving the British banks behind and returning to the U.S.

Here, many smaller regional and community banks use systems from FIS, Fiserv, or Jack Henry. Even modularized and opened up with APIs, these traditional vendors have systems originally created 10 to 25 years ago.

Back in 2015, Anthony Jabbour from FIS seemed relaxed commenting that smaller banks move to new systems at the same rate they always have — “about five per cent year.”

Banking has sped up since 2015 and this was before the FinTechs showed up.

Is the future FinTech?

Maybe. The U.S. is a unique banking market with a distinctive culture, so, the odds are against any FinTech thinking it’s remotely similar to Europe or Asia.

However, there are some interesting and modern choices from the newer FinTechs focused on core banking system replacement. There are many, but I took a look at three. One British, one European, and one from the U.S.

Thought Machine

Paul Taylor, CEO of Thought Machine, writes that in the “banking world, where it is common to find systems where no one has any real understanding of how components written years ago work anymore, nor any idea how they would be removed or replaced.”

To solve this problem, Thought Machine released VaultOS in 2016 to be “the engine for the banks of tomorrow.” It’s based on blockchain and machine learning technologies.

Their goals are ambitious. Thought Machine wants to “not only build a core banking engine in the cloud, but a full retail bank, with treasury systems, financial reporting, CRM, credit risk decisioning.” The Google connection is interesting. The CEO, CTO, CIO, COO, and many of their engineers, worked at Google before joining. But banking, especially U.S. banking, is difficult and different.

I have not seen this system, but they do have a strong product story. For now, Thought Machine appears to be focusing on the U.K.


Mambu’s headline implementation is with N.26, their Berlin neighbors. And N.26 is now arriving in the U.S. with a bank-in-an-app and minimalist design. Mambu is selling into the U.S. However, their challenge is adapting to the U.S. Boston or Boise are not Berlin.

One noteworthy implementation was for Oak North Bank, a challenger bank in the U.K.. Moving its core banking system from in-house to AWS, using Mambu's system, cut technology costs by 60%. 


A new core-as-a-service startup focused exclusively on the U.S. is Finxact. The U.S. banking system is unlike any other (4,774 banks is the starting difference). Finxact understands this deeply.

The founder and CEO of Frank Sanchez knows the industry and its technical oddities profoundly. In the early 1980s, Frank Sanchez and his brother Mike started Sanchez Computer Associates and created a core banking system called Profile. They sold their company to FIS in 2004. Frank Sanchez then ran the Enterprise Banking Division at FIS for eight years.

Finxact just raised $30 million from strategic investors.

Still do it

In sum, it still costs less to replace a core system than to try to compete in a digital world with legacy technology.

But, if you’re a bank and think traditionally about replacing core banking systems, then, you should plan to fail. Thinking rigorously about your capabilities and what can go wrong is more productive than unrealistic optimism.

All the modern software advice applies. Be agile. Start small. Build incrementally from prototype to MVP to minimum viable solution (changing humans is hardest).

And consider the newer FinTech vendors with state-of-the-art products and architectures.

If you have any questions, please Contact Me.

Consultative Selling Cleanup

Consultative Selling Cleanup