Financial services companies are investing in artificial intelligence faster than they are repairing the fragmented customer data systems needed to support it, creating a risk that personalisation programmes fail to improve loyalty or service.
Research from AND Digital found that approximately two-thirds of financial services organisations regarded fragmented data as a barrier to artificial intelligence driven loyalty. The study drew on responses from 250 enterprise level business leaders.
Disconnected systems limit the ability to create a consistent view of each customer across products, channels, and interactions. Some 61% of respondents said their organisation still struggled to produce a single customer view, while 68% feared losing customers during the next year if experience did not improve.
Cost was another constraint. Fifty-nine per cent said the technology required to support artificial intelligence powered loyalty was too expensive, and 72% believed effective personalisation depended on large scale data operations already being in place.
Most respondents recognised the importance of those foundations. Eighty-five per cent agreed that data was the most important factor in delivering a strong customer experience, yet 75% said their organisation prioritised artificial intelligence investment over repairing the infrastructure required to make those systems effective.
Kenn van Hauen, chief AI officer at AND Digital, said: “AI is reshaping how loyalty works in financial services. Larger organisations are already using real‑time data and AI to deliver faster and more personalised experiences, while many firms are still trying to bring fragmented systems together.”
He added: “Most organisations understand the value of more personalised customer experiences but delivering them depends on having reliable data and the ability to act on it. The organisations investing in those foundations today will be best placed to retain customers and stay competitive as customer expectations continue to rise.”
Fragmentation is particularly difficult in financial services because relationships often span systems built at different times. Current accounts, loans, insurance policies, investments, mortgages, credit cards, and support interactions may each use separate identifiers, consent records, and product histories.
Mergers and acquisitions add further complexity. Institutions can spend years integrating customer records after a transaction, while regulatory retention requirements and ageing core platforms limit how quickly information can be consolidated.
Third party distributors, brokers, payment providers, and outsourced service companies create additional boundaries. A customer may interact with several organisations during one journey, while no single system contains a complete or current record.
Artificial intelligence cannot compensate reliably for inconsistent source information. When addresses, contact preferences, balances, complaints, vulnerability indicators, or product holdings are incomplete or duplicated, personalisation can become irrelevant or harmful.
A recommendation intended to strengthen loyalty may instead reveal that the organisation does not understand the customer. Conflicting messages, inappropriate products, and repeated requests for information can undermine confidence more quickly when they are delivered automatically at scale.
The investment imbalance reflects different commercial incentives. A chatbot, recommendation engine, or generative assistant can be demonstrated quickly, while data architecture, governance, lineage, identity resolution, and consent management are less visible and often require coordination across several departments.
Connecting new tools to existing silos through temporary integrations can increase complexity without resolving the original problem. Each additional layer makes future migration more expensive and can make it harder to explain how a customer outcome was produced.
Evidence that artificial intelligence adoption is separating organisations with mature foundations from those struggling to scale is particularly relevant in financial services. Larger institutions may carry more legacy complexity, yet they also possess greater capital, larger data teams, and the transaction volumes needed to justify real-time systems.
Customer listening presents a related weakness. A failure to collect and act on client feedback can translate directly into lost revenue, while artificial intelligence can identify patterns only when the information is connected to the correct relationship and available quickly enough to influence action.
Financial institutions must balance personalisation with privacy, fairness, and regulatory obligations. A system using sensitive information to tailor an offer may produce a commercially attractive result while raising questions about consent, explainability, discrimination, or the treatment of vulnerable customers.
Data quality therefore requires sustained ownership rather than a one-off technology programme. Product teams, risk, compliance, marketing, customer service, and technology functions need common definitions, access controls, correction procedures, and clear responsibility for each source.
Investment decisions should distinguish between applications that can operate safely on existing information and those requiring deeper remediation first. Narrow, well governed use cases can produce measurable value while exposing weaknesses before systems are extended across the customer base.
The 59% concerned about cost face a choice between continued patching and structural change. Consolidating platforms is expensive and disruptive, although maintaining duplicated data, manual reconciliation, and inconsistent customer journeys creates recurring costs of its own.
Loyalty will ultimately be reflected in retention, complaint levels, service speed, product suitability, and trust rather than the sophistication of the model. Companies that continue placing new tools on fragmented foundations risk automating the inconsistencies customers already find frustrating.
You must be logged in to post a comment.