Out-of-date core technology is no longer sustainable in an era of real-time banking, as Dan Barnes discovers
UK banks have been pushed to update their core technology, following a string of IT failures and criticism from the regulator.
For a bank, changing its core system is like moving house. It is a rare event, it is complicated, risky and expensive. So rather than move from their 1960s and 1970s core platform “bungalows”, the bigger banks have preferred to build “extensions” on them to cope with change, until now.
Following an investigation of major banks’ technology, triggered by very public failures at RBS in 2012 and 2013, Sam Woods, a director in the Prudential Regulation Authority (PRA), reported the regulatory body discovered a number of deficiencies with the top eight UK banks and that “there is a programme of remediation under way”.
Speaking before a House of Commons committee in January, Mr Woods described banks’ technology as “antiquated”. According to PRA, a lack of investment in new technology has had consequences, including unreported failures across the other banks as well as the RBS systems breakdowns.
Although banks spend a huge amount of money on technology, it has always been around the edges of the main core, says Cormac Flanagan, product director for BankFusion Universal Banking at platform provider Misys.
“Banks have picked off everything to do something with, but the core,” he says. “Everything else is quicker to do, there is quicker payback. They have looked at channels, they have looked at payments, anything rather than the core systems.”
The core system holds all the information on customers and products, so any front-end activity on a mobile channel, for example a balance query or a payment, requires the mobile technology to connect with the core system and the customer’s account.
The effect of these Frankenstein-bank platforms has been to drive up costs, reduce reliability and impede customer service
That might sound easy, but the systems were designed in the 1960s and ‘70s when payments were processed overnight, rather than in real time. Accounts were held at bank branches, hence the need for a branch sort code. Customer records were linked to products, so a mortgage record and a loan record would be held separately. When ATMs were introduced in the 1980s, offering real-time payment and account checking, banks ran them alongside this old branch system, instead of replacing it.
“As far as I know, no bank running branch accounting has ever replaced it,” says John Schlesinger, chief enterprise architect at core system supplier Temenos. “In the 1990s and 2000s several banks tried and failed.”
For 30 years, banks have attached new technology to antiquated technology, building layers of middleware to manage records on mainframe computers that pre-date decimalisation in an era of real-time electronic currency. When they merged, banks would simply run the systems side by side.
Nick Brewer, product strategy director for banking at Misys, reports finding up to 20 core systems within a single bank.
“Five main functions might make up one core banking piece and then you have a merger of three or four banks, which have all had their platforms left running,” he says.
The effect of these Frankenstein-bank platforms has been to drive up costs, reduce reliability and impede customer service.
José Maria Fuster, chief information officer at Santander, says: “By storing transactions according to the product bought, the old core banking platforms had no single view of a customer. In the case of the UK, everything is also tied to the sort code of the branch, creating an additional constraint. If a customer bought a new product, the bank created a new record, linked to a branch, instead of simply adding the product to the existing customer record.”
Santander began migrating customers of Abbey, Alliance & Leicester and Bradford & Bingley on to its home-grown Parthenon platform in 2007. Although the transition was far from smooth, the end-effect has been to deliver a real-time customer service that offers a single view of the customer.
“Parthenon is customer-centric, so any transaction instantly updates the customer position so there is no need for heavy-duty overnight processing,” says Mr Fuster. Santander claims to make savings of £480 million a year by operating a single core system and has seen customer satisfaction continuously improve over the last three years.
Nationwide has also bought a new core platform; a five-year project resulted in a new mainframe-based core platform from software giant SAP being installed in 2012. New accounts from its current account range – FlexAccount, FlexDirect and FlexPlus – have been running on it since September 2012. Older customer accounts are held on Nationwide’s original Unisys platform and will be migrated across to SAP later in 2014.
Other banks may now be forced to follow suit. The introduction of a seven-day account switching procedure in September 2013 has seen traditional banks losing customers. The UK Payments Council reported 306,240 account switches in the fourth quarter of 2013 – 17 per cent up on the same period in 2012. The winner in this process of customers moving accounts has been Santander followed by Nationwide, while HSBC has been worst hit, followed by Lloyds. Improving service quality is paramount.
Dan Latimore, senior vice president at analyst Celent Banking, says: “With the introduction of mobile, customers are hitting the core more often for balance checks and they expect real-time information rather than wait overnight for the batch to run.”
However starting out on a six-year project is not easy, if no one in the bank has any expertise to call upon. Martin Vonk, general manager at the Competence Centre Core Banking for ING Bank, says: “It is hard to build and capture any experience in core banking migration projects. Complex by nature, often embedded in a specific home-grown legacy environment with multiple customisations and interfaces, these projects are executed with low frequency, so the repetitive learning is limited.”
ING developed a unit that specifically aimed at capturing these skills and experiences. As requirements can and will change overnight, he recommends that functions be composed of building blocks that you can take apart and put together, instead of putting all the functionality in one core engine.
“Business as well as IT architecture questions should be addressed upfront. Clean up the database, get rid of obsolete functionality and data, diminish the number of products, redesign and simplify processes,” he says. “Decide what is in, what is out in a core engine – the general ledger, regulatory reporting, CRM [customer relationship management] capabilities. As a result, a so-called logical application architecture and target operating model provide the guidelines or roadmaps behind any core-engine transformation.”