Change management in banking is the process of helping employees adopt new technology, workflows, and ways of working without losing trust, productivity, or institutional knowledge along the way. Most banks treat it as a training problem. It's actually a communication problem.
Gallup data confirms AI use at work is rising fast, with the share of employees who believe their organization has implemented AI climbing from 33% to 44% in a single year. But BCG found only 25% of frontline employees say they receive sufficient guidance from leadership on how to use it effectively. Banks don't have a resistance problem so much as a communication problem wearing a resistance costume.
Why resistance is rising, not falling
Surveys of banking organizations put resistance to change and talent availability as the two top barriers to implementing AI, cited by roughly two-thirds of both large and small institutions. Overcoming AI adoption barriers in banking consistently comes down to this same gap: change management isn't a soft-skills afterthought, it's a top-two blocker, on par with the technology itself.
The fear isn't irrational. DBS and Intesa Sanpaolo have both cited AI directly in recent layoff announcements, cutting 4,000 and 9,000 jobs respectively. Salesforce rebalanced its own support division from 9,000 employees to 5,000, with CEO Marc Benioff confirming AI agents now handle 50% of customer interactions.
Removing people without a change plan, however, doesn't just create resistance; it creates worse outcomes than the resistance was protecting against.
Change management in banking isn't one plan. It's three.
Most change management plans treat "employees" as one group. Banks run three distinct front lines, and each resists AI for a different reason. Call it the three-class method: identify which group you're managing change for, then match the response to the fear that group actually has, not the fear leadership assumes they have.
Customer support
The economics are the loudest signal in this group, and employees hear them clearly. When a company the size of Salesforce cuts support headcount by 44% and credits it to AI, every contact center employee at every bank reads that headline. The fear here is direct and specific: my job is the cheapest one to automate.
Klarna's reversal is the counter-argument worth using with this group. AI handled volume, but it couldn't handle the judgment calls that kept customers satisfied, which is exactly why the company brought humans back. The message that lands isn't "AI won't touch your role." It's "AI took the queue, not the judgment, and the bank that skipped the judgment had to rehire for it."
Operations
Operations resistance rarely sounds like fear. It sounds like exhaustion, and a McKinsey Global Institute study found employees spend nearly two hours a day searching for information scattered across tools. That's roughly 480 hours a year, or twelve full work weeks, spent searching instead of producing. McKinsey's own framing has stuck for over a decade for a reason: companies are effectively hiring five employees and getting four who show up for productive work.
Backbase has watched that number play out literally. At the company's Engage Americas 2026 keynote, CEO Jouk Pleiter described an operations employee who requested a third physical monitor because the applications required for the job no longer fit on two screens. He put a figure on the scale of it: 60% of operations work now lives in the whitespace between disconnected systems, not inside any one of them.
The fix for this group isn't a town hall about AI. It's proof that one specific, painful, manual handoff disappears.
Advisors and relationship managers
This group's resistance is the most misread, and there's now solid survey data to correct the read. Advisor360°'s 2026 Connected Wealth Report, surveying 300 US financial advisors, found 74% already say AI helps their practice, and 90% don't believe AI will make their role obsolete within the next decade. But 93% also say retaining control over decisions and advice is non-negotiable. Separate research from Edward Jones and Morning Consult, surveying advisors in May 2026, found 82% already using AI. 53% see it mainly as a chance to focus on higher-value client work, and 68% say long-term trust still requires a human touch.
That's the real shape of RM resistance: not "will this replace me," but "will this let me actually advise, or will it just watch me." Arun Ramamoorthy, Backbase's own Head of Commercial Banking, put the before-picture bluntly: RMs have become a "to-do list task master" for clients who call asking where their paperwork is. What the data shows, and what he expects, is the same shift: AI compiling the research so the RM spends the meeting advising instead of searching.
The leadership fix isn't confidence. It's closing the usage gap.
Most leadership advice on AI change management says "communicate clearly" or "be honest about uncertainty." That's not wrong, but it's not specific enough to act on.
Leaders use AI at nearly double the rate of the people they lead, and that gap is the real credibility problem. Axios reporting on Gallup's data found that in organizations offering AI tools, 67% of leaders use it daily or a few times a week, compared with just 46% of individual contributors. Most leaders are asking employees to trust a tool the leaders have already quietly mastered, without giving employees the same runway to get there. The fix isn't a confidence performance. It's closing that usage gap before asking anyone to follow.
Investment, not reassurance, is what actually changes how safe people feel. ADP Research's 2026 Global Workforce Survey of 39,000 workers found employees who feel their employer is investing in their skills are 5.3 times more likely to feel their job is secure. Among those who feel that investment, 53% are fully engaged. Among those who don't, only 12% are. Daily AI users are already more engaged than non-users, 30% versus 14%, which makes the point sharper: the fear isn't AI itself, it's feeling left to face it alone.
Where public honesty still matters: it's what makes the first two moves credible instead of performative. Jouk Pleiter modeled that on stage at Engage Americas: "I'll fully admit that I'm a little bit intimidated by it. At moments, I'm a little bit scared." That line lands because it's paired with action, not instead of it.
What actually closes the gap, by role
Generic AI training sessions don't work because they answer a question nobody in the room asked. Each role needs a different proof point, not a shared one.
For customer support, the gap closes when the AI visibly takes the queue, not the judgment. Salesforce didn't just cut headcount; it redeployed employees into professional services, sales, and customer success. That's the proof to show this group: routine volume goes to AI, escalations and judgment calls stay human, and the job doesn't disappear; it changes.
For operations, the gap closes with one dead handoff, not a transformation roadmap. Robert Soetens named the real blocker as comfort, not capability: "We've just become very comfortable with human beings filling in the gaps." The fix isn't a presentation about AI's potential. It's picking the one manual handoff this team complains about most, killing it, and letting them watch it stay dead.
For advisors and relationship managers, the gap closes with tools built for their specific job, not a generic copilot. JPMorgan Chase's rollout is the clearest proof of this: the bank didn't hand every employee the same tool. It built COiN for legal document analysis and CoachAI specifically for wealth management advisors, running on one shared platform used by roughly 250,000 employees. An RM shown a tool built for someone else's job doesn't trust it. An RM shown a tool that already knows their client's portfolio does.
The pattern across all three: nobody adopts a promise. They adopt evidence, scoped to the specific fear their group actually has.
Progressive modernization reduces the burden by design
Part of why change management in banking gets so hard is self-inflicted. Big-bang transformations force every employee to absorb every change at once. Progressive, journey-by-journey modernization spreads that disruption over time instead, so no single release becomes the moment everyone has to relearn their job.
That same logic runs through how Backbase frames the operational whitespace between systems: the goal isn't one dramatic AI launch, but a series of small, provable wins that build trust one workflow at a time. The Banking OS sits above existing systems for exactly this reason, so change lands as an addition employees can verify, not a replacement they're forced to trust blindly.
Frequently asked questions
What is change management in banking?
Change management in banking is the structured process of helping employees, customers, and operations adapt to new technology, workflows, or organizational shifts without losing trust or productivity. For AI adoption specifically, it means addressing the fears of customer support, operations, and relationship management teams differently, since each resists change for different reasons.
What causes employee resistance to digital transformation in banks?
Employee resistance in banking usually comes from three sources: fear of job loss, exhaustion from fragmented tooling, and a lack of clear communication from leadership about how AI will actually be used.
How is change management different for customer support vs. operations vs. relationship managers?
Customer support resistance is driven by visible industry layoffs tied directly to AI. Operations resistance comes from years of tool fatigue and manual handoffs between systems. Relationship manager resistance is more about identity, whether AI will reduce the role to data entry or free it up for real advisory work.
Do bank employees actually lose their jobs to AI?
Some do. Salesforce, Klarna, and several major banks have cited AI in real layoffs. But Klarna's own reversal after cutting support staff shows AI-driven cuts without a change plan can backfire, forcing companies to rehire for the judgment AI couldn't replace.
What's the first step in a bank's change management plan?
Start by removing one visible, painful manual task for one employee group, and prove it publicly. Trust builds from evidence, not from a training deck explaining what AI could theoretically do.
