Three in Four Bank Customers Are Switchable: What the Celent Study Means for How Banks Manage Relationships

Three in four bank customers are, at best, moderately satisfied with their primary bank. That is the headline figure from "The Banking Expectation Gap: Global Consumer Edition", a study Celent conducted for Temenos and published on 24 September 2026. Celent surveyed 2,515 banking customers across five regions between June and August 2026 and added its Dimensions survey of 216 banking leaders. Celent calls this group the "switchable middle": customers who do not leave, but would not stay either if a competitor offered better value, a better app or a product that fits their situation.

For banks this is not a satisfaction statistic. It is a question of how customer relationships are managed. Rarely has a study shown the contrast so plainly between what customers ask for and what banks prioritise: just 4 percent of the banks surveyed name personalisation of the customer experience as their top investment priority. At the same time, 56 percent say winning and retaining customers has become harder over the past year.

What the numbers say for Europe

The European results track the global picture closely, with one notable feature: dissatisfaction with payments is the single largest item here as well. 54 percent of European respondents cite payment frustrations as a source of dissatisfaction, 34 percent the digital experience. 71 percent are, at best, moderately satisfied with their primary bank.

Source: Temenos/Celent, Global Consumer Edition, 24 September 2026 and European results.

What stands out is how strong the expectation of anticipation has become. One customer in two expects the bank to already know what the visit might be about when they open the app or call. That is precisely the capability retail banking has traditionally been weakest at: the bank knows the balance, not the reason.

Switchable does not mean switching

The switchable middle is dangerous precisely because it does not show up in conventional metrics. These customers have not closed their accounts, they rarely complain, and they answer satisfaction surveys with "fine". One in four respondents worldwide has recently given serious thought to switching banks; in the US edition of the study, 62 percent of banks say acquiring and retaining customers has become more difficult.

The study also shows that dissatisfaction and switching triggers are not the same thing. Payments are the most frequent source of frustration, yet only a quarter of customers in the Middle East and Africa, where dissatisfaction is highest, would switch for that reason. Celent attributes this to inertia, with one exception: among 30- to 39-year-olds, one in three would change banks for better payment services. The real switching drivers lie elsewhere. In the US edition, 51 percent cite better rates and fees on credit products, 42 percent a more personalised fee or rewards structure, and 39 percent better online and mobile features; in Europe, 37 percent name the latter.

This is the gap competitors move into. A bank that rewards the relationship, offering terms that reflect how long and how broadly a customer has banked with it, meets an expectation around 40 percent of customers state explicitly. Most banks already have the instrument for this: the transaction and product data of their existing base. What is missing is the link between that data and a decision that reaches the customer before they start looking elsewhere. We described what that step looks like in practice when we examined McKinsey's numbers on AI personalisation.

AI: yes to guidance, no to autonomy

The study confirms a pattern other surveys have shown this year. 68 percent of customers worldwide would use a conversational interface for banking queries. Considerably fewer than half would "definitely use" AI features that make purchases on their behalf or execute routine payments. Younger cohorts, according to Celent, show particularly strong interest in AI-powered personal financial advice; the willingness to delegate actions is growing more slowly than the willingness to be given explanations.

The concerns are specific and more pronounced in Europe than in the global average: 49 percent cite privacy and data security, 40 percent errors and wrong decisions. This matches the Deloitte survey on bank customers' trust in generative AI we assessed a week ago. For the bank, the sequence is clear: first explain, contextualise and propose; only then, and only with explicit consent, execute.

What banks are planning

On the bank side, the Dimensions survey shows movement, but in a different place. 46 percent of banks plan major changes to their core banking systems in 2027, or a complete replacement. More than a fifth intend to extend AI initiatives beyond internal operations into direct customer contact. Both are sensible, but neither reaches the switchable middle by itself. A new core system changes the posting, not the conversation. And an AI project in customer contact without a decisioning logic creates new channels, not new relevance.

The study thus points to a gap between technology strategy and customer strategy. Asked what keeps customers, Celent answers trust, relevance and convenience. Only one of these three, convenience, is a direct result of system modernisation. Relevance comes from data and decisions; trust comes from explainability and from the experience that the bank does the right thing at the right moment.

What this means for managing the relationship

From a customer lifecycle management perspective, the switchable middle can be worked in four steps:

1. Identify. Willingness to switch shows in behavioural change, not in complaints: a falling number of transactions, a salary payment moved elsewhere, a card that stays unused after a failed payment. A learning retention loop rather than a static churn score picks up these signals while there is still time.

2. Prioritise. Not every moderately satisfied customer is equally valuable or equally at risk. Customer value and churn probability together determine who receives an intervention and who is merely monitored.

3. Reward the relationship. Terms that reflect the length and breadth of the relationship are an explicit customer wish, according to the study. They are also the instrument competitors find hardest to copy, because they presuppose the existing relationship.

4. Anticipate and explain. When the customer opens the app or calls, the bank should know the probable reason and make a proposal the customer can follow. This is the point where AI in customer contact builds trust instead of spending it.

Each of these steps is measurable, and each can be tested against a control group. That is what distinguishes customer steering from customer satisfaction programmes: it does not answer whether customers are satisfied, but whether a specific action caused a switchable customer to stay.

Five takeaways

1. 71 percent of European and nearly 75 percent of bank customers worldwide are at best moderately satisfied with their primary bank; one in four has recently considered switching.

2. Only 4 percent of banks name personalisation as their top investment priority, while 56 percent find acquiring and retaining customers harder than a year ago.

3. Dissatisfaction and switching triggers differ: payments frustrate most, but customers switch for pricing, relationship rewards and digital features.

4. Customers want AI that explains and guides; privacy (49 percent in Europe) and errors (40 percent) are the hurdles for AI that acts on its own.

5. Core banking modernisation does not reach the switchable middle by itself; what works is the link between behavioural signals, prioritisation and relationship-based offers.

Illustration: AI-generated. AI-assisted content: We use AI technologies and automated agents in the creation of our articles, including from Microsoft, Google, OpenAI, Anthropic and other providers. Topics, editorial direction and final approval remain with our team.