Somewhere over the past few years, the argument about personalised engagement in retail banking quietly stopped being an argument. Asked what will separate the successful banks over the next three years, the leaders who took part in the research Moneythor conducted with FinTech Futures put digital speed and convenience first, on 39%, with personalised engagement close behind at 20%. Competitive pricing and fees, which for most of the industry’s history would have led a list like that without much difficulty, came third on 16%.
Nor is this an industry sitting on its conclusions. Banks in the study are launching an average of sixteen personalised engagement initiatives a year, and taking somewhere around 4.8 months to move each one from concept to launch. That is a considerable amount of delivery for a capability supposedly stuck in pilot.
And yet the outcomes have not shifted in proportion to the effort. Customer drop-off still averages 13%, with 79% of banks reporting somewhere between 5% and 30%, against a relationship that generally takes more than twelve months to become profitable. Margin pressure and new customer acquisition were named as the biggest barriers to retail growth in equal measure, both on 22%. Read those findings alongside the delivery figures and the obvious explanations start to look thin. This is not a sector that lacks conviction about personalisation, and it is not one that has failed to invest.
What it does lack, on the evidence of the research, is control over when.
Two clocks, and only one of them is managed
Every customer relationship in a bank runs on two different clocks, and they have almost nothing to do with each other.
The first is the delivery clock, the one measured in months and quarters: the time it takes to scope a capability, get it approved, build it, and put it in front of customers. Banks understand this clock well. They staff against it, report on it, and have got measurably quicker at it, which is what those sixteen initiatives a year represent.
The second clock is shorter and much less visible. It starts when something financially meaningful happens in a customer’s life and it stops when the bank does something useful about it. A salary arrives a week later than usual. A streaming subscription renews at nearly twice last year’s price. A balance drifts towards a standing order that is going to clear on Thursday whether the money is there or not. A card is used abroad for the first time. Each of those is a moment where a bank could be genuinely helpful, and each has a window that closes.
Relevance is decided on the second clock. Most institutions manage only the first, which is why a campaign signed off in March cannot say anything sensible about something that happens on a Tuesday afternoon in June, no matter how well the campaign itself was built.
Why the weakest capabilities cluster the way they do
When respondents were asked where they are weakest against modern customer expectations, three areas came back: consistent omnichannel experiences on 30%, relevant and personalised engagement on 23%, and proactive financial guidance on 22%. In most banks these sit with different owners, on different roadmaps, with different budgets.
They are closer to being one problem than three. Omnichannel consistency, looked at properly, has little to do with channel coverage, since most banks are present everywhere their customers are; it has to do with whether every channel knows the same thing at the same moment. Relevance is rarely a content problem either. The content usually exists. What is missing is a reliable way of knowing which piece of it matters to this customer today rather than in the quarter when the segment was drawn. As for proactivity, it is almost entirely a statement about timing. Guidance that reaches a customer after they have already worked the situation out is not guidance in any meaningful sense, whatever the quality of the writing.
Personalisation has widened faster than it has quickened
The personalisation figures in the research make this visible in a way the narrative findings do not. An average of 38% of customer interactions are now personalised using AI or advanced analytics, which sounds encouraging until the distribution underneath comes into view.
- 48% of banks deliver targeted offers or messaging
- 29% provide context-aware interactions at key customer moments
- 8% deliver highly personalised journeys across most touchpoints
- 6% have reached fully individualised, real-time experiences
Moving down that list, personalisation becomes progressively more timely and progressively rarer. Targeted messaging is a decision taken in advance about a group. Context-awareness requires knowing something about where a customer is at the point of contact. Individualisation in real time means deciding inside the moment itself, with no opportunity to review the decision first.
Coverage was always going to be the easier axis to move, and moving it first was a reasonable sequence rather than a failure of nerve. It does, though, have a ceiling that arrives sooner than most roadmaps assume.
AI adoption has followed much the same logic. It sits today largely in back-office automation, fraud detection, anti-money laundering monitoring, and support chatbots, all functions where the clock is forgiving and the loop stays inside the bank. Only 8% of institutions regard themselves as market leaders in AI adoption, and on average respondents estimate they are running about 1.9 years behind the innovation curve. Customers, meanwhile, have not agreed to wait: 42% already expect AI-driven interactions from the digital channels their bank provides.
The risk calculation has quietly reversed
Two thirds of respondents, 67%, now say that moving too slowly to meet customer expectations is a greater risk than choosing the wrong technology. A decade ago the anxiety ran firmly the other way, towards the danger of selecting badly and living with it for years.
That reversal makes more sense once both clocks are visible. If the binding constraint is coverage, then a measured pace is defensible, because coverage accumulates and a year’s delay costs roughly a year’s progress. If the constraint is timing, the arithmetic changes, since every moment that passes without a useful response is a relationship that did not deepen, and those moments do not queue up waiting to be answered later.
It also casts a different light on a finding that is easy to read as complacency. Some 54% of respondents believe they can improve their retail customer metrics over the next three years without fundamentally changing their digital engagement model, while 65% expect customers to switch banks within two years if expectations go unmet. Held together, those two positions are perfectly coherent for a bank whose problem is coverage. For a bank whose problem is timing, they are not, because no amount of extending a batch process produces a real-time decision. Which of the two describes a given institution is worth establishing deliberately rather than discovering later.
What sits between a financial event and a useful response
For banks that have closed some of this distance, the work has been less about spending more than about shortening the gap between something happening and something being said about it.
Enrichment does the unglamorous part. A raw transaction line is close to meaningless until it has been resolved into a recognised merchant, a category, a recurring subscription, or a salary credit, and everything downstream inherits the quality of that resolution. It is also the reason platforms built for clicks and demographics tend to struggle in banking; they were never designed to read financial behaviour. Decisioning then determines what is worth saying to this particular customer now, on the basis of what has just happened in the account rather than the segment they were placed in last quarter. Orchestration puts it into the channel they are already using while the moment is still open.
Taken together, that is what Moneythor means by Deep Banking: personalisation treated as a baseline rather than a differentiator, proactivity that anticipates rather than responds, and value that reaches past the transaction into rewards, referrals, and loyalty.
The research is clear that banks want this governed rather than autonomous, which is the right instinct in a regulated industry. Some 30% are comfortable with AI orchestrating customer journeys inside pre-approved guardrails and 27% with AI recommending next best actions for a human to approve. Delivery preferences have settled along similar lines, with 53% reporting that third-party implementations have historically produced the most successful outcomes, and integration flexibility ranked as the leading criterion when choosing a partner, on 28%.
Where the layer is in place, the outcomes are measurable. Moneythor clients report an average 25% increase in active customers and up to 2.5 times more in savings deposits. Trust Bank in Singapore captured 12% of the Singapore retail banking market in under a year, and 70% of its customers have been acquired through referrals. At Chiba Bank, 79% of users who opened a campaign message went on to complete the full journey. As the research notes, modern platforms of this kind use API-based integration to add capability without replacing core banking infrastructure.
Get the free executive summary
Closing the Gap: Customer Expectations, AI Strategy and Personalisation in Retail Banking sets out what retail banking leaders told Moneythor and FinTech Futures about engagement, personalisation, and AI, and where the distance between ambition and delivery currently sits.
Free Download