Data rich, insight poor: the personalisation paradox in bancassurance

Data, process and recommendation logic still rarely combine into a single operating model. Personalisation in bancassurance continues to function at a surface level, despite banks holding a stronger starting position than most other channels. Three areas stand out.

First, data does not translate into decisions. Information about the customer exists but is not used consistently. Most institutions lack a mechanism that genuinely connects product, behavioural and contextual data with the recommendation process. Second, there is no single recommendation logic. Separate channels, separate campaigns and separate priorities make consistency and scale difficult to achieve without centralised decision-making. Third, process overrides the recommendation. In many banks, needs analysis remains a formal confirmation exercise rather than a real personalisation tool – its existence does not yet translate into better protection matching.

The requirement to conduct a customer needs analysis is firmly embedded in Europe's insurance distribution regulatory landscape. The Insurance Distribution Directive and the national legislation implementing it placed an obligation on distributors that was meant to address something far more important than a form: the connection between a customer's need and the protection being proposed.

After years of living with these requirements, this is precisely where the market continues to lose most of its value. Supervisory authorities across Europe have highlighted not only cases where needs analysis is skipped entirely, but also situations where it is superficial – conducted hastily or reduced to obtaining confirmation that it has been completed. Common deficiencies include analyses based solely on data already held by the distributor, conversations limited to generic questions that can hardly underpin a meaningful recommendation, and analysis shaped around the product rather than the customer's full underlying need. Particularly telling is the case where the conversation focuses on securing a credit obligation rather than identifying which product would genuinely suit the customer.

The broader a bank's product range becomes, the more these limitations show. With a growing number of variants and standalone products, simply completing a needs analysis no longer confers any meaningful advantage. The process must be carried out because regulation requires it – but without stronger recommendation logic, the bank gains proportionally little in personalisation, product fit or business value. Treated not as a compliance cost but as a starting point, needs analysis can become a genuine source of better recommendations and greater value for both the customer and the bank.

Effective personalisation cannot rely solely on a form or a single process step. It requires an architecture in which technology connects customer data, relationship context, the sales moment and recommendation logic within one coherent model. As product ranges and contact channels multiply, data integration, rules engines, decisioning capabilities and cross-channel consistency all become critical.

Four foundations matter most:

• A single point of decision-making – a central recommendation mechanism operating across channels – ensures consistency and selects the next best action regardless of where the customer engages.
• Connecting data to the decision requires an architecture that links relational, transactional and behavioural data to the recommendation process in real time; this is where AI models play a growing role, not as a substitute for business judgement but as support in assessing context and prioritising offers.
• Explaining the recommendation matters because an effective output must provide the adviser and customer with a clear rationale – explainability, audit trails and reproducible decision logic are becoming as important as the models themselves.
• Finally, designing the journey, not just the offer means choosing the right moment, channel and form of presentation, not only the right product.

This is the approach underlying RIFE, our bancassurance platform. The recommendation mechanism is designed as an independent layer, separated from other processes, allowing recommendation logic, customer journeys and product catalogues to be managed and evolved independently. A bank can begin with a simple needs analysis form and develop it progressively into a more advanced model without rebuilding the entire sales process.

A growing number of institutions are building recommendations as a separate decision layer – one that brings order to the product range, supports the adviser and makes better use of available data.

At Rabobank, a central engine gathers signals from physical and digital channels, identifies customer intent and selects the next most relevant interaction – moving away from segment-and-campaign thinking towards a more individual view. The bank reports 1.5 billion personalised interactions per year, a four-fold CTR increase and a 208% improvement in marketing conversion. At a major UK bank, a real-time customer data platform and decisioning layer shortened the path from data to action dramatically, enabling consistent recommendations across mobile, online, branch and call centre – with a 300% increase in loan sales among mobile users and a 90% reduction in data-to-action lag. At Crédit Agricole, the recommendation tool does not replace the adviser but structures the conversation: narrowing cover options, supporting the argument and guiding product selection. Results included a 50% uplift among insurance specialists and 200% among bank advisers. At National Australia Bank, personalisation was built as a single central decision mechanism across channels, ensuring the bank makes one consistent decision based on current customer context rather than channel-specific logic.

Beyond the bank's own data, open finance is increasingly providing the additional context recommendations need. In Brazil, open finance has evolved into a practical extension of customer knowledge, with around 85 million active data-sharing consents – meaning personalisation starts not with additional questions but with a fuller picture of the customer's financial life.

The shared conclusion is simple: personalisation works when data, the moment of contact and recommendation logic combine in a single operating model. The broader the offer, the more the decision layer, data integration and cross-channel consistency matter.

Bancassurance is entering a phase where simply having an insurance proposition is no longer a differentiator. In a world of broad product ranges, growing standalone offerings, rising customer expectations and an increasingly rich data ecosystem, the institutions that will win are those that can recommend and personalise more effectively. That requires a shift in mindset: needs analysis should be seen not as a regulatory box to tick but as an integral part of a broader personalisation model. The future of bancassurance will depend not on the number of products in the range, but on the quality of the decisions a bank is able to make on behalf of – and together with – its customers.