Banks have spent heavily on AI but most of it is still chatbots and back office work — not tools that help banks grow. Getting new checking customers now costs more than twice what it did in 2018. And when banks merge, some customers take their money and leave. Joining me is Sid Singh, CEO of Curinos. Sid, thanks for being with us today.
Yeah, thank you.
How much of AI in banking is actually making decisions today and how much is still a test run?
Bank CEOs are telling me — we are not buying tokens, we want to buy business outcomes. Banks are evolving from using AI in the back office and simple task automation into customer decisioning oriented around outcomes. That means AI in deposit decisioning — who is likely to open an account, who is likely to fund balances. We are seeing some green shoots, but still very early days.
What should banks be asking to make sure they are getting a real return on AI investment?
There is huge pressure on banks to drive organic growth. Cost of acquisition has doubled. Consumer behaviour has changed — agents are now looking for bank deals on their behalf. It is no longer a consumer shopping for rates. It is a consumer's agent. How does a bank compete in that era? Banks have to get sophisticated in agent-oriented behaviour. Having data on your consumer becomes even more important than ever before.
Getting a new checking customer now costs 126% more than in 2018. Why?
Two trends. First, large national banks are winning more market share because of their scale. Smaller fintechs are nipping away at the lower end. Regional banks are squeezed in the middle — not enough size, not enough data, not enough technology. Second, consumers are more aggressive in switching accounts. There are about 20 million switchers — people likely to move their bank accounts. Inside that population, four out of every switcher has at least four accounts. Cost of acquisition has seen a big increase and the ROI is not panning out. Banks have to become more sophisticated in finding deposits inside their existing consumer base rather than always prospecting new customers.
When banks merge, why do customers leave and how do banks keep them?
Our data shows that above a certain M&A deal size — $5 billion and above — 11% of deposits can actually leave. Doing due diligence not just on credit quality but on deposit quality is incredibly important. Customers leave because integration sometimes does not go as well as planned, communication is not as clear, and the product proposition is not as joined up. Banks need to focus on deposit quality, the product proposition, and a real integration experience. One plus one should equal eleven.
You helped a bank bring in $1.6 billion in new deposits without offering higher rates. How?
That is an example of authentic AI. Banks have been fairly sophisticated in deposit pricing — how to offer the right price to the right customer. But there is still opportunity. AI is moving banks into deposit decisioning — which customer do I market to, who is likely to move their account, who is likely to fund a new account with balances? By combining the bank's own data, Curinos data, and third-party signals, we can build best-in-class data models that drive $1.6 billion in new deposits — not a single rate offer, all incremental.
What does the Databricks partnership mean for banks trying to put their data to work?
If everyone uses the same AI models, everyone becomes mediocre. The differentiator is data. Our partnership with Databricks allows us to take data from different parts of a bank — even those still using legacy infrastructure — through a layer that ingests it and shortens the path to outcomes from six, nine, or twelve months to literally a few weeks. Banks can now see AI-led value in days.
What are you seeing in Middle East bank consolidation?
There is continued pressure on banks to drive deposits. Higher interest rates are a mixed blessing — you can drive loan growth, but there is pressure to offer higher deposit rates to keep low-cost deposits. We expect the same consolidation trend to continue — not to accelerate significantly, but to continue both in the US and the Middle East.
Thank you so much for joining us.
Yeah, thank you for having me.
