Banks Know Their Customers. Why Is It So Hard to Turn That Knowledge into Growth?
By Nicole Volpe
Simple Subscribe
Subscribe Now!
Over the past two decades, financial institutions have steadily expanded the customer and member insights they have at their disposal, advancing from demographic profiles and customer segmentation to enriched transaction and behavioral data. In theory, armed with such capabilities, every institution should be highly skilled at deepening and expanding accountholder relationships.
Reality check: But that’s not the case.
The reasons why are complicated. Institutions struggle to translate customer intelligence into business results both because of how they are organized and how they go to market. To break through, they must not only adapt to AI but reinvent how they approach customer relationships more broadly.
To be sure, the expansion of generative and agentic AI in financial services offers a way forward for institutions looking to get better at “connecting the dots” by bringing accuracy and speed to product design and offer generation.
Why this matters: The rise of consumer-facing AI agents also raises the pressure on traditional institutions, which must manage new risks even as they act with urgency to stay ahead of competitors. Meeting these challenges will require them to make fundamental changes in the strategies and tactics they bring to a fast-changing marketplace. To learn more, The Financial Brand spoke with Carson Kotnyek, Head of GTM and Advisory at Zafin.
Ensure Alignment on All Sides
An important first step for most institutions is to look inward. Banks often have siloed organizational structures, typically built up over years or decades, that make it harder for them to optimize customer or member relationships. Each line of business has its own technology platform and go-to-market strategy. And no one in the organization is truly able, or incentivized, to manage the customer relationship as a whole.
Kotnyek sees the incentive problem as two-sided. “Most of the time, line of business incentives do not ladder up to a shared metric of success for the bank,” he said. “So they end up driving conflicting incentives for the customer that don’t serve the bank’s interest.
As a result, an individual mortgage-holder who has a business checking account and a credit card with the same bank might receive communications and offers that reflect different goals and incentives. In some cases, branding might vary from one account to the next. All of this can lead to confusion and missed opportunities, as conflicting messages and disjointed campaigns crowd accountholders’ mailboxes, with the net effect of reducing engagement and trust.
Why this matters: These conditions also undermine growth and retention, as institutions fail to capitalize on the most basic customer insights. For example, savings account transaction data may reveal an accountholder’s intent to buy a new home, but the team overseeing core banking may have little incentive to coordinate with the mortgage unit on a response that would protect the broader relationship.
The good news: Institutions do not need to rip and replace multiple product platforms to remedy the situation. They can start with a coordinated strategy, including aligned performance goals that measure total relationship value (e.g., total asset growth, retention rates, or cross-sales).
Create Full-Relationship Value Propositions
Another effective way to solve the silo problem is by building relationship programs that operate across business lines and provide a benefit and incentive structure that aligns the customer’s goals with their institution’s goals.
Loyalty and membership programs, mostly linked to credit card spending, have a long history in financial services as effective drivers of engagement. More than 60% of U.S. card issuers and fintechs surveyed by PYMNTS Intelligence said rewards programs increased total spending and 66% reported increased card use. At the same time, top-performing loyalty programs can increase annual revenue from customers who redeem rewards by 15% to 25%, according to a McKinsey analysis of loyalty programs across industries.
More financial institutions are starting to see broad-based relationship programs as a way forward, where deeper relationships unlock progressively richer benefits. “Banks are starting to see the potential,” Kotnyek said. “They’re thinking about loyalty as not just credit card points but as a holistic program.” Large institutions and fintechs already have such programs in market:
- Bank of America’s BofA Rewards — originally launched as Preferred Rewards in 2014 and relaunched this year — provides benefit tiers on a customer’s combined bank-deposit and Merrill investment assets. At the top two tiers, benefits extend beyond banking to include credits for streaming and news subscriptions, events, and experiences.
- PNC’s TotalRewards, introduced in 2026, similarly offers membership tiers based on balances across banking and investment. As with BofA, higher balances and additional products unlock progressively richer rewards — including higher savings rates, enhanced card rewards, and cash incentives for qualifying mortgages and auto loans.
- Revolut’s subscription model (launched in Europe in 2017 and the U.S. in 2020) lets customers purchase benefit packages at different levels. These combine financial advantages, such as higher savings rates, reduced international transfer fees, and expanded ATM withdrawal allowances, with lifestyle benefits including third-party subscriptions and travel perks.
Program design can also reward customers for staying. Kotnyek described a bank offering a 1% match on a $100,000 investment transfer and paying the $1,000 out monthly over 12 months. By the time the customer has collected the full reward, the relationship is much more likely to last.
Prepare for Agent-Mediated Competition
The rapid adoption of personal AI agents that can execute financial transactions — including Meta’s mass-market Muse and Spear Street Technology’s invitation-only Instinct — has presented financial institutions with a range of new risks and opportunities.
Some institutions are responding defensively, focusing on security concerns and the potential for agents to accelerate erosion of banking relationships by helping consumers more efficiently pursue higher rates or lower fees. “Blocking agents due to security risk is totally valid but it’s not going to be a long-term answer,” Kotnyek said. He cited small- to midsized businesses seeking operating efficiencies with payroll and fund transfer agents. “Will these businesses stay with a bank that blocks their agents from logging into their account?”
Why this matters: Institutions with relationship programs will be especially well positioned to retain customers and their funds. But they must make program rules and customer-specific benefits accessible to authorized AI agents in machine-readable form, so agents empowered to optimize can “see” the full picture and assess the full impact of moving money elsewhere.
A Checklist for the Future
As banks increasingly use AI to translate customer intelligence into personalized offers and actions, governance becomes essential to ensuring those actions are executed within clearly defined policies, permissions, and approval processes. This is particularly important as agents gain access to sensitive customer data and greater authority to initiate actions. A new article by Zafin offers a checklist that can help financial institutions keep their feet on the ground as they reinvent their go-to-market strategies for AI.
- Establish each agent’s identity and authority.
- Enforce policies and controls during execution.
- Govern agents consistently across operating environments.
- Route consequential decisions to humans with full context.
- Monitor and control the cost of agent operations.
- Document every agent action and decision for auditability.
- Manage the agent lifecycle and prevent uncontrolled proliferation.
A handful of core principles underlie this checklist:
- Governance should be applied to the workflow, not simply the agent.
- Limits of agents must be enforceable, in real time.
- Human authority must be built into the process.
- Audit trails must be generated as work happens. This is especially critical when it comes to offer generation, Kotnyek said: Institutions must be able to explain pricing changes and why similar customers receive different offers.
Bottom line: A financial institution’s ability to maximize its potential in the age of AI will be a function of the quality of its governance. An agent gathering customer intelligence might operate independently with read-only access to customer data, while one making pricing changes could require human approval before those changes take effect. As agents demonstrate reliability, institutions can expand their authority within defined limits. This gives banks a way to automate increasingly complex and consequential work while maintaining control over decisions that affect customers and the institution.
