Banks: it's time to break free of your legacy tech
You don’t have to spend much time in retail banking before you come across legacy systems - a catch-all term often used to explain why banks are so slow and unresponsive.
We’ve lived for almost a couple of decades with ubiquitous and universal internet access. We’ve moved from desktops to mobile and, on-premise, to the cloud. The sophistication of consumer and business serving systems is remarkable. Every business now has access to technology that enables them to create, innovate and ultimately better engage customers. Banks, too, have made pretty impressive progress. Their apps have improved incrementally over the years and now contain all sorts of features that make banking easier for everyone.
Neobanks like Monzo and Starling raised the bar and have driven a lot of progress over the last five years. Incumbent banks knew that customers wanted more, but until someone else actually provided it, there was safety in numbers for mediocre experiences. The improvements in customer experience made by incumbents have been massive. When you think about the technology backdrop against which these changes were made, it’s frankly nothing short of miraculous. Delivering features in the legacy-rich incumbent banking environment is hard, risky, slow and very, very expensive.
The can is at the end of the road for incumbents
Delivering the same features from a neobank technology stack isn’t trivial or easy but it’s significantly faster and much less expensive per customer. Neobanks tend to come in at about £20 per customer, incumbents closer to £150. So far, incumbents have managed to get away with throwing money at the problem. Their massive customer base produces economies of scale that allow them to afford bloated technology budgets. Only a fraction of this is spent on genuine innovation with the majority used to ‘keep the lights on’ and generally maintain the IT estate. The net result is a banking experience that works but hardly inspires. The pound per pixel cost of any banking app is huge, but banks point to the fact that they’ve got decent uptime, comply with substantive regulatory requirements and are constantly evolving.
Only a fraction of this is spent on genuine innovation with the majority used to ‘keep the lights on’ and generally maintain the IT estate.
Every time a new feature or service is added, the estate acquires an additional layer of complexity. This adds costs to the future. In its own right, this model isn’t sustainable in the long run. And it gets worse. Neobanks continue to make inroads into incumbent bank customer bases. It’s not quite one-for-one as a lot of customers still hold accounts with both, but the erosion of the customer base is certainly an issue in the medium term. It keeps getting worse for the banks when you consider their overall revenue model. Tech entrants like Apple have already hoovered up revenue from interchange which makes up about 30% of retail banking revenues. This is just one example of fast-developing trends in consumer financial services.
Up and down the banking stack there is a tonne of (often VC-funded) innovation taking place. Start-ups are looking to capture market share from every aspect of the banking value chain. From clearing to payments to embedded finance and so on. While there are strong arguments for seeing the total financial services market as growing rapidly, it's fair to say that for every successful start-up, the revenue gained is revenue lost for incumbent banks. This is a whole other world to explore, but suffice to say banks are starting to feel the pinch. With their revenues under threat, they need to innovate. The problem is, that this tends to inflate their cost base, so what options do they have? They can cut staff, automate processes, adopt scaled agile processes or use more AI, and some of these things may well help (or more likely not if our experience is anything to go by). But in the end, all roads lead to legacy.
...for every successful start-up, the revenue gained is revenue lost for incumbent banks.
At their heart or ‘core’, banks run on technology that was developed over 40 years ago. There have been upgrades and improvements over time, but ultimately mainframe computing remains the bedrock of banking technology. Originally designed as a digitised ledger, these systems have, over the decades, had to accommodate increasingly complex use cases that they simply weren’t designed to support in the first place. They’re incredibly efficient at some things but generally inflexible. This old, expensive and rigid technology stands in stark contrast to what the neobanks and fintechs are able to deploy.
We’ll skip the usual buzzwords, but the net result is that businesses running on internet-scaled technology are able to innovate faster, better and more efficiently. Any number of metrics can be used to illustrate the difference, but my favourite is the number of technology releases to production that a bank is capable of. Incumbents used to aim for once a quarter per platform. They have now improved to once a month. A neobank can release tens of times a day. Every day. This is possible for a number of reasons, ranging from delivery methodology to culture, but they all rest on the simple fact that new technology performs better than legacy.
So, what options do banks have?
While evolving the core has worked reasonably well to date, it's hard to see the sustainability of the model when fintechs and Banking as a Service providers get up a head of steam. Quite a few have managed to ‘virtualise’ their legacy cores by effectively copying the system from mainframe infrastructure to more modernised virtual machines. This has benefits from a cost-to-run perspective but sustains the critical rigidity that inhibits the bank’s ability to innovate. In other words, it buys time but little else.
Another approach is the so-called ‘Lifeboat’ strategy. In this model, the theory is that a bank builds a brand new banking stack similar to those used by the neobanks. It comes with all the potential advantages of efficiency and flexibility. The idea, then, is to migrate customers from the legacy system onto the new ‘greenfield’ platform where banks could operate on par with new market entrants. Great plan. Sadly, it isn’t that simple.
The regulatory environment requires the deployment of significant risk management strategies to minimise any downtime and ensure that customers are able to migrate over in a non-disruptive manner. Such a migration is hard enough when the two systems are broadly similar. Hard, but certainly not beyond the reach of well-drilled banking IT teams. The process is risky but comes with a lot of use cases within the risk tolerance of the bank. Having to deal with ‘edge cases’ or features and capabilities that have accrued over time on the legacy platform poses a problem. These can range from simple exception handling for certain customer segments to more complex joint account linking, and so on. Again, much of this complexity can be managed, but with increased risk and cost.
The real trouble comes when such edge cases are not documented or implemented in code buried deep inside the system. Over decades, the number of these types of issues has increased and become obscured. This results in the Rumsfeldian ‘known unknown’ problem. When considered together, these issues eventually mean one thing. Banks can’t be sure that if they move customers over to a new platform everything will work as it did before. And that is a killer. There are ways around this and scope can be managed to mitigate the risk (e.g. migrate simple savings accounts only), but ultimately it means you’re only solving part of the problem.
So we’re basically left with one option. There are lots of metaphors used to describe replacing core banking systems. A brain transplant is a personal favourite. For those of a more mechanistic persuasion, we have ‘switching out the plane engines while flying’. Whichever your choice, one thing is for certain - it’s risky and expensive. Risky and expensive enough to deter pretty much the entire industry C-suite to date. But this sentiment is clearly changing.
Look at the number of ‘core banking’ solutions now available in the market. Temenos has done a good job of keeping itself relevant in the world of ‘cloud-native, microservices, NoSQL, etc.’ with its Transact offering. Mambu and Thought Machine are established brands, offering brand new, component-oriented tech stacks that are starting to make sales. More broadly, there still seems to be plenty of VC money in the space. Of course, sales are crucial for any business, but in core banking, this is exponentially important. No bank wants to be the first to take on a risky venture. But after one successful implementation, the rest will follow. To date, we’ve only been able to identify sales on the periphery of the bank’s estate. Savings accounts and sole trader business accounts with low complexity and ultimately low risk of contagion should it all blow up. So this type of implementation might provide some proof points or evidence that the technology works but, frankly, we knew that already.
No bank wants to be the first to take on a risky venture. But after one successful implementation, the rest will follow.
Are there any options left for a bank exec with genuine regard for the long-term value of the bank’s shareholders? The tenure of a typical bank CEO is around three years. Any core banking replacement programme is likely to take a similar amount of time to execute, let alone procure. So it’s easy to see why most top bank execs aren’t motivated to make the call. This leaves most banks floating about in a place similar to that occupied by the slowly boiled frog.
Where to go from here
There are ways to replace core banking in a modular way (i.e. not ‘big bang’). One is to replace a horizontal layer with new technology. At 11:FS, we call this the ‘brownfield horizontal’. The other is to start with one product on new (vertical) infrastructure and then move across the bank to more complex areas. We call this ‘brownfield, thin slice’. A brownfield horizontal strategy has its share of merits, but it’s mostly a tech change with limited upside for the customer. The brownfield, thin slice strategy can provide more radical change for its customers, and comes with the added benefit of giving the bank an opportunity to change the overall operating model to behave more like a fintech. In the long run, you can evolve this strategy by taking further thin slices into the new domain and spreading both the cultural and technical benefits of the new approach. In fact, in the long run, this feels like the only genuine option.
For more fascinating insights like these, be sure to check out our ‘Rebuilding financial services from the inside’ report - it's a belter.