Study

AI in Banking Apps: A Study on Blocked Payments and Chatbot Failure

A consumer study, measuring two failure points in how Europe's most-used banking and payment apps handle problems: how often users have a payment, transfer, or transaction blocked, declined, or paused by their app, and how often the chatbot they're sent to afterwards fails to resolve it.

Consumers don't trust AI with their money

Just 5.4% trust a chatbot to fix problems. 65.2% want a human. A 12-to-1 preference

Chatbots fail nine times out of ten

Only 11.4% of blocked-payment users say the chatbot actually resolved their issue

Half of blocked payments never happen

48% of affected users say the transaction died and they had to find another way to pay

Chatbot complaints are surging across Europe

Negative reviews blaming chatbots rose 55.49% year over year across Europe's 10 biggest money apps

What we studied

This study looks at two specific failure points in how banking and payment apps handle problems for their customers: how often a payment, transfer, or transaction is blocked, declined, or paused by the app, and how often the chatbot the customer is sent to afterwards fails to resolve it. We refer to these two failure points together as the block-and-bot experience, because in practice they are rarely separate events. One leads directly into the other.

The reason for studying them together is that each failure point is usually measured in isolation, if at all. Fintechs report fraud-block rates internally without measuring downstream resolution. Chatbot vendors report deflection rates without measuring whether the customer's problem was actually solved. By measuring both in the same study, using a consumer survey for the inside view and two years of European app store reviews for the outside view, we can show what the combined failure looks like from the customer's side, and how the pattern is moving across the market.

The Chatbot Failure

76.3% of users with a blocked payment ended up dealing with a chatbot

Even when consumers try to avoid AI customer service, they end up at it anyway. Of the 1,162 affected users, 56.6% chose in-app human chat as their first resort and only 12.0% picked the chatbot. But by the end of resolution, 76.3% had dealt with a chatbot at some point, three in four. Only 19.2% managed to speak to a human throughout.

The chatbot resolved the problem just 11.4% of the time

Of users routed through a chatbot during a blocked-payment resolution, only 11.4% say it actually solved their problem. Half (50.0%) had to escalate to a human, and 14.9% never got a real resolution at all. The chatbot's job is to deflect work away from human agents. The data shows it deflects roughly one in nine cases successfully, and forces a second contact for the rest.

Explore the press release and visual highlights

Get the official press release and view key data visualized in infographics designed for media use and sharing.

View press release and infographics
The App Store Evidence

Chatbot-blame negative reviews rose 55.49% year over year across Europe's 10 biggest money apps

A scrape of 159,600 reviews across the 10 most-used banking, wallet, and payment apps in Europe — Revolut, bunq, Monzo, Wise, N26, Sumeria, Starling, Bourso, Monese, and Chase UK — found 34,167 negative reviews (1-star and 2-star) in the analysis window. Reviews blaming a chatbot or AI assistant for an unresolved problem rose 55.49% year over year. The trust gap isn't a survey artefact. It's pan-European and growing.

Identity verification and declined transactions are the top problems chatbots can't resolve

Among chatbot-blame negative reviews where the underlying problem was classifiable, the four most common unresolved categories are identity verification (16.85%), declined transactions (16.09%), login and access issues (12.74%), and account suspension (11.23%). Together they account for 57% of all classified chatbot-blame reviews. These are the exact failure modes that need a human, and the exact failure modes a well-built digital banking platform is meant to prevent reaching the customer in the first place: identity edge cases, account-level decisions, and time-sensitive payment problems.

The Trust Gap

Users trust a human agent 12 times more than a chatbot to fix problems with their money

Asked who they trust most to actually resolve a problem with a banking, wallet, or payment app, 65.2% of consumers pick a human agent. Just 5.4% pick the chatbot or AI assistant, a 12-to-1 preference for humans. Another 19.1% trust neither and expect to have to escalate, and 10.3% say they don't know. Taken together, that's nearly one in four app users (24.5%) who have no faith that the first thing they touch when something goes wrong will actually help them.

Of users who've been blocked, half have been blocked more than once

Of the 77.2% of surveyed app users who have had a payment, transfer, or transaction blocked, declined, or paused, half (50.4% within the affected base) say it has happened more than once. Within the full sample, 38.9% have been blocked repeatedly and 38.2% only once. Almost an even split. The block isn't a one-off error pattern that hits a user, gets resolved, and goes away. It recurs.

The Block Experience

77.2% of surveyed app users have had a payment blocked, declined, or paused

More than three quarters of surveyed banking, wallet, and payment app users say their app has blocked, declined, or paused a payment, transfer, or transaction they were actually trying to make. The experience is split almost evenly between people it has happened to once (38.2%) and people it has happened to more than once (38.9%). Only 21% say it has never happened to them. Behind each of those decisions sits a payment acquiring stack making a real-time call on whether to let the transaction through.

Only 13.4% of users consciously feel their app is over-cautious, a 64-point gap with reality

When asked directly whether their app is too cautious with their money, only 13.4% of users say yes (1.7% often, 11.8% occasionally). Nearly half (48.7%) say they're not sure, and 37.9% say no. Yet 77.2% of the same sample have had a payment blocked. The 64-point gap between conscious perception and actual experience suggests users treat each block as a one-off rather than as evidence of a wider pattern.

The Cost of the Block

56.7% lose hours or days, and nearly half see the transaction never complete

Of the 1,162 users whose payment was blocked, 56.7% say resolution took hours or days. For 48.0%, the transaction never went through at all. They had to find another way to pay, or didn't pay. The block doesn't just inconvenience the user. Half the time, it kills the payment entirely.

10% used a different app or bank afterwards, the churn signal

When a payment is blocked, 10.0% of affected users complete it using a different app or bank. Another 27.6% are embarrassed at the point of failure, often in front of someone at a checkout, and 15.0% miss a bill or incur a late fee. One in ten lost transactions becomes one in ten lost customers, at least for that purchase, and potentially for good.

Explore the press release and visual highlights

Get the official press release and view key data visualized in infographics designed for media use and sharing.

View press release and infographics

Final thoughts

The story behind the data is simple. People get blocked. They get to a chatbot. The chatbot doesn’t fix their problem. They waste their time (and money in some instances). All of this gets stored in their memory the next time they must choose which application to pay with.

This is because fintech companies have spent the last few years slashing costs by incorporating AI into their companies. Fewer humans to deal with customer issues mean more chatbots and stricter rules on what transactions are allowed. While these solutions work well for the company, they create a worse experience for consumers when something goes wrong.

The 55.49% rise in chatbot complaints across Europe has resulted in customers leaving one-and-two-star reviews for these companies, publicly naming the chatbots as the cause of their issues. This is bad for the brand and the company’s capacity to retain customers. It is also drawing the attention of regulatory authorities.

The question is no longer whether fintech companies should use AI to handle customer issues but where they should use AI in their companies and what aspects of the customer experience they should allow AI to decide.

The answer is human-centric fintech companies that automate the nonessential aspects of customer experience and allow humans to resolve challenging customer issues. This is the way the fintech industry will have to develop in response to the consumers’ experiences.

Methodology

The study combines an analysis of negative app store reviews across Europe's most-used money apps with a UK consumer survey.

Two years of app store review analysis: apify scrapers pulled 1- and 2-star reviews from the Apple App Store and Google Play for the 10 most-used digital banking and payment apps in Europe: Revolut, bunq, Monzo, Wise, N26, Sumeria, Starling Bank, Bourso, Monese, and Chase UK. 159,600 total reviews analysed, of which 34,167 were negative. Each negative review went through two classification passes: first to identify chatbot or AI-blame sentiment, then to tag the underlying problem the user was trying to resolve. The period 1 April 2025 to 30 April 2026 was compared against 1 April 2024 to 30 April 2025 to surface trend movement. Results are reported in aggregate across all 10 apps; no app-by-app grading is published.

1,506-respondent UK consumer survey: fielded between 27 April and 4 May 2026 to UK adults aged 18+ who use at least one major banking or payment app and had used it within the past month. Age and location quotas applied to match the UK adult population. Eight questions in total, two qualifying and six main, with conditional logic routing the resolution and impact questions only to respondents who had experienced a blocked payment (n=1,162).

For press and editorial use

The findings on this page are free to cite and reproduce with attribution to DECTA. Please reference the study as "DECTA, AI in Banking Apps: A Study on Blocked Payments and Chatbot Failure (2026)" and link to https://www.decta.com/. For interview requests, additional data cuts, or comment from DECTA, contact marketing@decta.com.