How To Build a Data Foundation for the AI Era
JUNE 17, 2025
By: Tyler Brown
More than half of respondents to a KPMG survey (54%) said their bank has “implemented” foundational data capabilities. But few claim data capabilities that are “fully developed and operational,” and in the age of AI, the last step is increasingly out of reach. As we’ve written, the more data banks have, and the more effectively they process and use data, the more competitive they will be. But some don’t even have basic capabilities designed to fix siloed, inconsistently formatted information.
To succeed in the long run, a bank’s leadership will need a clear vision of the future of data as it applies to their institution and the industry overall. And, to make that vision a reality, they will need a data strategy that’s linked to enabling technology and business practices. In our research, we’ve observed that successful data strategies follow three principles:
Data is a strategic asset that’s crucial to decision-making.
Effective governance is fundamental to data quality and compliance.
Systems that create and process data should be seamlessly interconnected.
Applying these principles, banks can begin to think about infrastructure. Based on our research and experience, forward-thinking data strategies typically start with established tools and look ahead to the mainstreaming of AI across the bank. There are three main tiers to be aware of when building a data stack:
Application layer: Contains customer- or employee-facing applications such as the general ledger, ERP, and CRM; loan origination and operations; digital banking; payment systems, and risk and compliance tools. All applications create data, which is frequently interrelated, but the data they produce may be in different formats, have different characteristics, and be difficult to reconcile.
Processing layer: Collects, stores, and transmits data from applications. A data layer will include a data warehouse, data lake, or combination of the two, to allow data that’s created across the bank to be queried. It will also include data extraction software that pulls and cleans data from disparate systems. Event-driven architecture and modern APIs enable the latest generation of data management systems to stream data between applications.
Intelligence layer: This is where AI models are developed, trained, or validated. As we’ve also noted, AI models depend on high volumes of data, ideally made available to them in real time. Models then need to be deployed and monitored for compliance. The intelligence layer dovetails with the application layer, to provide real-time access to the bank’s data.
With this stack in mind, banks may adopt a roadmap for building out their data capabilities:
Awareness: Understand the value of holistic data to a bank’s performance, what may limit data use, and the organization’s need for technology that allows data to be used efficiently and effectively.
Vision: Articulate the types of data needed to inform business planning and identify the technology that will support the creation, collection, and seamless access to that data for all who need it.
Foundation: Roll out technology in line with the bank’s data strategy; create a data governance framework such that data from different systems can be used effectively in business processes across the organization.
Scale: Reach the point at which technology is integrated such that finding and using data is no longer a chore, because the data layer has been embedded into the bank’s workflows and access enables advanced analytics.
It’s important to remember that this is a long-term effort. The first step is to adopt the right mentality: to see data as a durable advantage and aim for decision-making driven by holistic, up-to-date insights. Tactical considerations regarding technology will follow and should be aligned strategically to the bank’s needs and resources.
Chime’s IPO: A Blueprint for Your Bank’s Next Decade
By: Kate Drew
June 18, 2025
It’s no secret that the neobank threat waned in the last few years. Once a major topic of conversation in strategic planning sessions, talk of such competition slowly faded among bankers I spoke with as these fintechs struggled to reach profitability and many shuttered their operations entirely. But now that discourse is back, for one very big reason: Unicorn neobank Chime went public.
Chime’s IPO reignited a conversation about the threat from fintech challengers to traditional banks. The debate is contentious — some see Chime as spend-happy with an unsustainable cost structure, while others see a pioneer. Either way, it is hard to ignore the company completely. More than that, it would be unwise to. In Chime’s public debut last Thursday, it reached a market cap of $14.5 billion.
Many people have written about why Chime matters from a competitive standpoint (and it does) but, more than that, it is part of a movement to change customer expectations in banking. The company built its name on a lack of branches, no fees, and slick experiences that drove its member numbers to 8.6 million as of March 2025. And, per the company’s S-1, 67% of those are in a primary account relationship with Chime. As such, I’d like to talk not about why we should fear Chime but about what we can learn from it.
Here are three points that stand out to me for traditional institutions:
Create value where it makes sense. There is so much emphasis put on Chime’s wellness products: early wage access, credit building, automatic savings roundups, etc. But what often gets lost is the why of it all. Chime is building financial wellness products because it focuses on customer segments that need those products. That strategy will not always translate. The lesson here is not to run out and add roundup functionality; the lesson is to take a Jobs To Be Done approach to your customers and build differentiated experiences based on their unique needs.
Think creatively about the business. By the business, I mean the business of banking. Alex Johnson of Fintech Takes wrote a piece recently about the importance of having a durable non-interest income strategy. Essentially, his argument is that because price optimization is only increasingly on the customer’s side, net interest margin is no longer a sustainable path to profitability. I happen to agree. Chime is a provider making money differently — we can argue about whether its interchange-based approach is wise, but it is the mindset that bankers should take note of.
Focus on transparency. One of Chime’s most obvious differentiators is its lack of fees — no minimum balance requirements, no overdraft fees, no foreign transaction fees, etc. But I’d argue this is really about transparency. Customers don’t want to be charged fees unexpectedly or for things they don’t understand. A fully no-fee structure is nice, but it probably goes beyond that. Chime has likely been able to use this initiative to build trust in an area that many consider a pain point. It also puts a great deal of effort into making its offerings relatable and easy to understand, including through its blog, an area where many banks could improve.
A common thread among these takeaways is they speak to a thoughtfulness that is missing in a lot of board rooms today. One of the arguments I often hear in defense of the status quo is that it is too hard to innovate, because of resources, because of regulation. That defense leapfrogs over strategic themes (like the ones above) to tactical hurdles. As bank executives contemplate Chime — and the fintech industry more broadly — they should examine approaches more than offerings and embrace discussion accordingly.
Ultimately, I am going to refrain from commenting on Chime’s future success. Not because I don’t have an opinion on it, but because in the context of a bank’s strategy, I don’t think it really matters. Competition is always going to be tough, and it’s always going to come from new and unexpected places. What matters more is how that competition changes the game and how the industry responds to it. For bankers, step one is to pay close attention. Step two is to find opportunities to meet the moment.