I was listening to a group of CIOs talking about legacy systems change the other day. One said that it doesn’t matter how old the systems are; it’s all about whether they are maintainable and fit-for purpose. I kind of agreed with that, although I cannot think of any silo-based, product-focused, fifty-year-old, batch-based system that could be in any way fit-for-purpose today.
Another said that COBOL is perfectly fine as a programming system as it is easy to learn and use, and is still appropriate for some systems today. Sure, maybe for the ones I just mentioned above.
A third said that core systems change is very hard and risky, as illustrated by TSB. Yes it can be, if you do a complete systems swap-out over a weekend with no roll-back plan, which I wouldn’t recommend.
Finally, a fourth said that, having changed core systems twice for two different banks, that he wouldn’t recommend it. His exact words were: “you spend three years changing the systems, during which time the bank is paralysed as you cannot change anything else; then, after three years, you get exactly what you had before on a new system that cost you a fortune and could have brought down the bank due to the risk”. He obviously had a bad experience.
I personally do not believe any of this baloney, as it’s a mindset from the last century when yes, core systems change was massively risky and costly. Today, with microservices architecture, open banking, 1000s of FinTech start-up firms doing one thing brilliantly well, and the ease of using plug-and-play software, I don’t think core systems change is hard at all.
I wouldn’t recommend a complete systems swap-out but a bite by bite movement. After all, how do you eat an elephant? One bite at a time. So, using conversion services of which there are many (Leveris, Solaris, BUD and so on), start converting the bank’s core systems one process, one product, one function at a time. Build a vision of where you want to get to, and start getting there. And one of the key aspects of the vision being built is that it has to be customer, not product, focused. And this is where I get into the main reason driving core systems change. It is nothing to do with code or legacy or risk. It is to do with intelligence. No large, traditional bank is ever going to be able to do customer-intelligent marketing and service if their customer data is fragmented and decentralised. Therefore, in order to get to use machine learning for customer intelligent marketing and service, a bank must change core systems over the next few years or face tough competition who does do this.
I blogged years ago about if Amazon behaved like a bank:
The book division doesn’t talk to the music division. The music division won’t talk with the electronics division. The retail business won’t talk with the wholesale business. The wholesale business won’t talk with the cloud business. Oh, and the Kindle division won’t even talk with the book division! None of them will share customer information with each other and, as a result, no-one knows what customers buy from Amazon as a group, when or with what sort of payment type. There’s no view on Amazon’s share of wallet or who is cross-selling what to each customer.
And I still believe this. This is the main reason why banks need cultural change and leadership from the top-down to get digital, as they have to break apart these old baronial product empires and start acting as an intelligent enterprise together.
That’s hard, but I’ve seen it being done and for those who don’t get it done, well, they just won’t be able to compete or survive. Just a thought.