That assumption is now being automated away. Roughly $23 trillion of the $70 trillion held in global consumer banking sits in accounts paying close to nothing, according to McKinsey’s 2025 Global Banking Annual Review. Global banking profit could potentially shrink by about $170 billion, or 9%, once third-party AI agents begin shopping rates and moving money on customers’ behalf. Consumer deposits and card lending absorb the steepest declines in that scenario, since both have depended most on customers not paying close attention. An agent that tells a customer they are leaving $2,000 a year on the table, and can act on it in a few taps, does to switching costs what mobile deposit did to branch traffic.

 

Many believe this is a big bank problem. However, community banks and credit unions have built their deposit franchises on relationships, proximity, and the trust that comes from knowing a customer by name, all of which lose their protective power the moment software makes the comparison and acts on it without a conversation.

 

None of this waits on agentic AI to mature, because the same erosion is already running through payments. Fintech competitors stopped attacking checking accounts head-on some time ago, entering instead through debit rewards, buy now pay later, peer-to-peer transfers, and wallets, then assembling those pieces into apps or platforms where customers actually manage money. J.D. Power found measurable attrition at incumbent institutions from exactly that pattern.

 

What makes it difficult to see from inside the institution is that nobody closes an account. Analysts call the behavior soft switching. In fact, a 2026 survey found one in five account holders had moved money away from their primary financial institution within the previous three months. The account stays open, the statement still carries the institution’s name, and the balance, the transaction volume, and the engagement have all gone somewhere else.

 

Faced with softening balances, most institutions reach for price, which is the one lever that reliably makes the underlying problem worse. Competing on rate against a fintech with no branch network and no legacy costs compresses margin without touching the reason the customer drifted in the first place: the other platform was faster and easier to use. A better rate does not repair an experience gap, and it does not win back the interface where the next money decision gets made.

 

That interface experience is what the competition is actually for, and it narrows the strategic question to one a management team can answer in a single meeting: Can we be the place where a customer opens the next account, applies for the next loan, and moves the next dollar?

 

Answering yes depends on a handful of operational capabilities, most of which have little to do with marketing.

 

The first is speed to open, measured honestly, not by the best-case number in a brochure, but by the median time from the start of an application to a fully funded account, including manual reviews, document collection, and any required follow-up. Customers have grown accustomed to opening financial accounts in minutes. When opening a small-business checking account at a local institution takes 11 days and requires a branch visit, the institution may lose that relationship before it even realizes the customer is considering other options.

 

Speed at the front-end matters less if the second account is as hard to get as the first. Continuity across products means a customer who opens a deposit account can apply for a loan without re-entering identity information, re-uploading documents, or restarting verification, because every point where a customer must repeat themselves is a point where they can leave. That continuity also produces something institutions rarely have, which is deposit, lending, wealth management, treasury management, and risk visible in one place at the same time. Pricing and retention decisions depend on precisely that view, and institutions running a separate system for each function end up reassembling it by hand, weeks after the moment it would have been useful.

 

Strategic and welcome use of AI in workflows is the third capability, and banks that prioritize the streamlining of processes will rarely forfeit new relationships and retention. Both consumers and business owners expect secure, AI-aided workflows in place of manual processes. Retaining deposits depends on financial institutions’ adaptation to those products and processes that prevent agents from moving the accounts or, more commonly, soft switching.

 

The final capability is the ability to adapt quickly. When a competitor introduces a product or feature that customers value, financial institutions need to be able to respond without spending months implementing new technology. Institutions that require several quarters to introduce comparable capabilities risk falling further behind customer expectations and the competitors already meeting them.

 

Most institutions struggle to deliver those capabilities because their technology systems are outdated and disconnected, not because they lack the right strategy. Account opening runs on one vendor, KYC/ KYB on another, consumer lending somewhere else, business lending on yet another, and treasury services on a platform that predates all of it. Staff move between systems to complete a single application,

and every handoff adds delay, rekeyed data, and another opportunity for the applicant to abandon. The customer experiences all of that as slowness, without ever knowing what caused it.

 

Consolidating that architecture is what closes the gap. When onboarding deposits, lending, wealth management, and treasury management with risk decisioning on a single platform of record, an institution can open an account, verify a business, originate a loan, and service the relationship without passing work between disconnected systems. The efficiency gain is the easiest part to measure, and the competitive effect is the part that compounds, since an institution able to act in minutes stays in front of the customer at the moment a decision is being made.

 

 

None of this requires a strategic planning cycle to assess. How long does it truly take, start to finish, for a new customer to open your most common account and have it funded? How many separate systems does a staff member touch to complete one business account opening? And when an existing customer wants a second product, how much of what you already know about them do you ask for again?