AI in banking

Retail banks make customers feel known again in 2027

16 November 2026
6
mins read

A decade of efficiency left retail customers feeling less known. How conversational AI moves banks from deflection to resolution in 2027, with data and what to watch.

The Backbase 2027 Banking Predictions Report maps 10 shifts for the year ahead: five that cut across the industry and five specific to a banking segment. This is the first of the segment predictions, and it’s about retail banking: how a decade of efficiency left customers feeling less known, and how conversational banking can win that relationship back. Below is the full chapter, with the data, the industry view, and what to watch.

TL;DR

  • Only 32% of consumers say retail banking service quality has improved in the last five years, and money is following them out the door.
  • Branch closures cut costs but removed many of the moments where relationships were built.
  • First-wave chatbots were built to deflect calls, so customers are three times more likely to use third-party generative AI tools than company chatbots.
  • In 2027, conversational AI helps banks know the customer, coach them, and act for them, moving from deflection to resolution.
  • Guardrails matter: 83% of customers would feel anxious about an AI agent acting on their finances without approval.

Faster, cheaper, and less known

Retail banking has spent the last decade getting faster, cheaper, and always-on.

As service moved from branches to self-service screens, banks cut the cost of every transaction and lost much of what made customers feel connected to them. Customers have noticed: only 32% of consumers say retail banking service quality has improved in the last five years.1 And their money is following. The average U.S. retail checking customer now holds deposit accounts at three different institutions, and one in five has moved money away from their primary bank in the past three months.2 In optimizing for efficiency, banks have put one of their most valuable assets at risk: deposits that stay. Banks have more data on their customers than ever, yet customers have rarely felt less known. The branch network shows how that happened. In the EU alone, the number of bank offices has declined by 2.62% since Q4 of 2024, and the number of bank employees by 0.80%.3 This shift made sense on cost, but it removed many of the moments where relationships were built. Customers feel that gap most when it matters: a transaction dispute, a scam, a major life event. When they can’t solve an issue online, 64% still turn to a branch to resolve it.1 Without these moments, relationships continue to leak, product by product.

Banks may still hold the account, but others increasingly own the relationship, and much of that churn is self-inflicted. For years, banks have used customer data mainly to push products, and 46% of customers say they feel pressured, at least some of the time, to accept products that serve the bank more than themselves.1 The first wave of AI chatbots widened the gap. Many banks built them to deflect calls, collecting a request and routing it elsewhere, so the customer’s real need often went unresolved. So customers are taking their questions elsewhere: they’re three times more likely to use third-party generative AI tools than company chatbots.4 In 2027, the banks that win customers back will use conversational banking to resolve what they ask, and deepen the relationship along the way.

β€œAI will make the experience more conversational. Customers will no longer be keen on going through different screens and filling in lots of forms. They will just talk to the AI assistant and say what they want: no more forms, no more screen by screen. So far, we’ve talked about AI helping the internal workforce; now, it’s time for AI to face our customers.”
‍
– Khiem Tran, Digital Banking Director, Techcombank

The industry view

In 2027, conversational AI will become how retail banks make their customers feel known again. It will help deliver personalization and empathy at scale: learning about customers’ lives (know me), guiding them towards their goals (coach me), and acting for them with consent and proper guardrails (act for me).

The biggest shift here is from deflection to resolution. Agents must have the ability to finish tasks within the scope of the conversation, without handoff or external mediation. The proof points are already in: DBS’s generative AI assistants now reach 10 million customers, and in H1 2026, they resolved 9 in 10 queries fully digitally, with calls and emails falling 7% and satisfaction rising 17%.5 Meanwhile, NatWest’s Cora handled 12.9 million conversations in 2025 alone, and the bank is currently testing voice-to-voice with humanlike empathy, tone, and inflection.6 Leading banks are already well beyond pilots and test cases, and financial institutions will need to lean into conversational AI if they want to remain competitive in 2027 and beyond.

Customer operations is where empathy will be proven first. Disputes, service requests, and account changes are where trust is won or lost, because the outcome is measurable and failure is visible. When done well, AI in bank customer care can cut calls by 25-40%, simultaneously lifting first-call resolution 15-25% and raising CSAT by 10-15 points.7 But keep in mind that the gains come from redesigning the work itself, end to end, rather than just slapping AI onto a fragmented process.

1.7x

Banks with the highest advocacy scores grew revenue significantly faster than the lowest, and advocates hold 17% more products with their primary bank.

Source: Accenture

However, conversational banking must also come with guardrails: consent, transparency, and a human in the loop when it matters most. Luckily, 62% of customers trust their primary bank to offer generative AI financial services vs. 19% for big tech.7 Yet 83% also said they’d feel anxious about an AI agent acting on their finances without approval.8 The burden of proof lies with the banks, both from the customer perspective, but more importantly from a regulatory one. Regulators and watchdogs across markets are turning broad AI principles into concrete expectations: how decisions are explained, how customer data is protected, and when a human will step in. In 2027, banks that build those safeguards into every conversation from day one will find that compliance and empathy pull in the same direction, and their customers will certainly notice.

Here’s what to watch in 2027

  • Banks start reporting resolution rates and satisfaction from AI conversations
  • Voice agents handle real service conversations, end to end
  • Household views and shared goals appear in mainstream banking apps
  • Human handoff promoted as a feature
Backbase 2027 Banking Predictions Report
Experts weigh in on industry-wide and segment-specific trends, from AI in production to banking by conversation. Get the full version as a PDF you can send to your team.
Download the report

Frequently asked questions

Why do retail customers feel less known by their bank?

As service moved from branches to self-service screens, banks cut the cost of every transaction and lost many of the moments where relationships were built. Only 32% of consumers say service quality has improved in the last five years, and 46% feel pressured to accept products that serve the bank more than themselves.

What’s the difference between deflection and resolution?

Many first-wave chatbots were built to deflect calls, collecting a request and routing it elsewhere, so the customer’s real need often went unresolved. Resolution means the agent can finish the task within the conversation, without handoff. DBS’s generative AI assistants resolved 9 in 10 queries fully digitally in H1 2026.

What results can conversational AI deliver in customer care?

When done well, AI in bank customer care can cut calls by 25-40%, lift first-call resolution by 15-25%, and raise CSAT by 10-15 points. The gains come from redesigning the work end to end, not from adding AI to a fragmented process.

Do customers trust their bank to use AI?

62% of customers trust their primary bank to offer generative AI financial services, against 19% for big tech. But 83% would feel anxious about an AI agent acting on their finances without approval, which is why consent, transparency, and a human in the loop matter.

Sources1. Accenture · 2. J.D. Power · 3. European Central Bank · 4. Gartner · 5. DBS · 6. NatWest · 7. McKinsey & Company · 8. Deloitte

About the author
Table of contents
https://www.backbase.com/banking-predictions-report-2027