An ISO/IEC27001:2013 and ISO 27018:2019 certified cloud solution
80 Key Customer Loyalty Stats & Trends to Watch in 2026 | Perx
Blogs
Beyond Loyalty: Why BFSI's $60B AI Opportunity Demands a Revenue Intelligence Layer
The investment is real. The ambition is clear. And the allocation is almost entirely wrong.
The global AI in BFSI market is projected to grow from USD 24.31 billion in 2025 to USD 60.09 billion by 2031 — a compound annual growth rate of 16.28% (ResearchAndMarkets, 2025). BFSI institutions across APAC and beyond are committing budget, headcount, and board-level attention to AI at a scale the industry has not seen since the shift to mobile banking.
But look at where that investment is actually landing. Fraud detection. Credit underwriting. Risk modelling. Back-office process automation. Anti-money-laundering systems. These are legitimate applications and meaningful improvements — and they are capturing the overwhelming majority of AI investment in financial services.
Meanwhile, the layer of the business that most directly determines whether a customer stays, grows, and chooses your institution over a competitor tomorrow morning is receiving a fraction of that attention. The customer-facing intelligence layer — the system responsible for understanding individual behaviour, driving engagement, and connecting those activities to revenue — remains the last unreformed frontier in BFSI’s AI transformation.
This piece examines why that gap exists, what it is costing institutions in real commercial terms, and what the architecture of the missing layer actually looks like.
TL;DR – Quick Summary
- BFSI’s AI investment is real but misallocated. The market is projected to grow from USD 24.31 billion in 2025 to USD 60.09 billion by 2031 (ResearchAndMarkets, 2025), yet most of it flows into fraud, underwriting, risk, and back-office automation.
- The customer-facing engagement layer, which most directly determines whether a customer stays, grows, and chooses your institution, is the last unreformed frontier in BFSI’s AI transformation.
- The gap is about measurement, not value. Fraud AI proves ROI in weeks; engagement AI needs measurement infrastructure most institutions have not built, so the best-funded applications are the most measurable rather than the most valuable.
- The cost is commercial: personalisation on transaction data alone misfires at scale, over 40% of consumers cannot distinguish between financial brands, and nearly 3 in 4 bank with more than one provider (UserTesting, Digital Banking Trends 2026).
- What is missing is a revenue intelligence layer spanning six connected functions, from data unification to revenue attribution. The urgency is sharpest in APAC, and with 53% of financial services firms already running AI agents in production, mostly in risk and fraud (ResearchAndMarkets, 2025), the engagement layer is open white space where first-mover advantage compounds.
Where the $60B Is Going — and Where It Is Not
The AI investment story in BFSI follows a predictable pattern: the applications that receive the most funding are the ones closest to regulatory mandate and the ones where the ROI case is easiest to make to a risk committee.
Fraud detection AI can demonstrate value in weeks: blocked transactions, recovered losses, reduced false positives. Credit underwriting AI reduces default rates and improves portfolio health — outcomes that map directly to balance sheet performance. Anti-money-laundering AI reduces compliance risk and the cost of regulatory penalty. These are defensible, measurable, and familiar to the board.
Customer engagement AI is harder to defend in those terms — not because the value is smaller, but because most BFSI institutions do not yet have the measurement infrastructure to prove it. They cannot connect a behavioural nudge to an incremental deposit. They cannot attribute a cross-sell event to a specific engagement campaign. They cannot calculate the revenue cost of a customer who became inactive over three months of irrelevant communications.
The applications that receive AI investment are not necessarily the most valuable. They are the most measurable — and measurement infrastructure is itself an investment BFSI has not yet made in the engagement layer.
The result is a lopsided AI transformation. Institutions are deploying sophisticated intelligence to decide whether to lend to a customer — and basic, rule-driven automation to decide whether to engage with them.
$60B
Projected AI in BFSI market size by 2031
ResearchAndMarkets, 2025
53%
of financial services firms deploying AI agents in production
ResearchAndMarkets, 2025
42%
of compliance leaders cite regulatory uncertainty as AI blocker
ResearchAndMarkets, 2025
The Commercial Cost of the Engagement Intelligence Gap
The absence of a genuine intelligence layer in customer engagement is not a theoretical problem. It has specific, quantifiable commercial consequences — most of which BFSI institutions are currently absorbing as the baseline cost of running a loyalty programme.
1
Personalisation without intelligence produces the wrong outcomes at scale.
Most BFSI engagement systems operate on transaction data alone. They know what a customer purchased and when. They do not know what product the customer holds, whether their engagement with the mobile app is declining, what their likelihood of cross-sell is, or whether they are three weeks from churning. Personalisation built on that partial picture does not just fail to help — it actively erodes trust when customers receive recommendations that are demonstrably irrelevant to their actual relationship with the institution.
2
Differentiation has collapsed precisely because engagement is underpowered.
More than 40% of banking consumers report they cannot meaningfully distinguish between financial brands (UserTesting, Digital Banking Trends 2026). This is not primarily a product problem — most institutions offer broadly similar deposit rates, card products, and digital features. It is an engagement problem. The experience of being a customer at Bank A versus Bank B feels, to most customers, functionally identical. An intelligence layer that understands individual behaviour and responds to it relevantly is the only mechanism capable of creating genuine perceived differentiation at scale.
3
The multi-provider reality is a direct result of engagement failure.
Nearly three-quarters of consumers bank with more than one provider simultaneously (UserTesting, 2026). This is the market telling BFSI institutions that none of them has yet built an engagement experience compelling enough to consolidate the relationship. When a customer holds their salary account at one institution, their credit card at another, and their savings at a third, it is not because the first institution lacked a loyalty programme. It is because no institution gave them a reason to consolidate — and that reason, when it exists, is always an experience, not a product.
4
Reward spend is growing without a corresponding growth in retention.
The default response to engagement failure in BFSI has been to increase the generosity of the rewards programme — higher earn rates, broader redemption catalogues, welcome bonuses. This approach has a structural ceiling: it trains customers to respond to incentives rather than to value the relationship, and it creates a cost base that grows proportionally with the customer base without building the compounding loyalty effect that justifies the investment. A digital bank that maintained a 72% returning customer rate did so through automated progression loops and behavioural mechanics — not through reward spend escalation.
These consequences share a root cause: the engagement layer is operating without the intelligence infrastructure that every other part of the BFSI business now takes for granted.
What the Missing Layer Actually Is
The phrase ‘customer intelligence’ has been used loosely enough in BFSI that it has lost most of its precision. CRM vendors call their segmentation tools customer intelligence. CDP providers describe their data unification as customer intelligence. The term has become a marketing category rather than a functional description.
The intelligence layer that is genuinely missing from most BFSI engagement stacks is something more specific. It is a system that performs six connected functions — and the absence of any one of them breaks the value chain.
Capability
What It Does — and Why Its Absence Has a Cost
Data Unification
The system ingests and standardises data from multiple sources: transaction feeds, product holdings, digital engagement signals, consent and compliance data, rewards interactions, and channel behaviour. Not a selection of these — all of them. Partial data produces partial intelligence, and partial intelligence produces the misfired personalisation that most BFSI customers experience today.
Customer Intelligence
On top of unified data, the system builds a behavioural profile for each customer: what their engagement trajectory looks like, whether their digital activity is increasing or declining, what their propensity for specific products is, and where they sit in their relationship lifecycle with the institution. This is not segmentation. It is individual-level intelligence that changes in real time as behaviour changes.
Growth Strategy
The intelligence layer translates customer profiles into growth recommendations. For a customer whose deposit balance has been static for six months and whose app engagement has dropped by 40%, the system identifies the intervention — the right mechanic, the right message, the right moment. It does not wait for a campaign team to identify the segment and brief a creative agency. It generates and prioritises the recommendation.
Execution Without Friction
Strategy without execution velocity is a planning document. The layer must be able to deploy campaigns, journeys, and behavioural mechanics without IT involvement for every change. In APAC markets where a challenger bank can push a new engagement feature to ten million users in a day, the institution operating on quarterly release cycles is not competing on the same terms — and no intelligence layer compensates for an execution bottleneck.
Experience Delivery
At the individual level, the intelligence layer delivers the right experience through the right channel at the right moment. Not a campaign to a segment — an interaction with a person. This is the difference between a customer receiving a push notification that is relevant to their actual situation and one that is relevant to their demographic cohort.
Revenue Attribution
The layer closes the loop: it connects every engagement action, every nudge, every campaign, every mechanic, to a measurable revenue outcome. Not a proxy metric. Not an engagement rate. Actual incremental revenue generated, churn prevented, cross-sell velocity accelerated. Without this, the engagement programme cannot defend its budget to a CFO — and in most institutions today, it cannot.
The institutions building this six-function layer today are not ahead of the market. They are building the capability that the market will require of every institution within five years.
Why This Matters More in APAC Than Anywhere Else
The global AI in BFSI investment story is relevant everywhere — but the urgency of the engagement intelligence gap is more acute in APAC for three specific reasons.
Market velocity is higher. APAC’s digital banking markets operate at a speed that has no equivalent in Western financial services. Mobile banking penetration exceeds 90% in several markets. Challenger banks have built multi-million-user bases in years rather than decades. Customer expectations, set by the superapps that dominate daily life in the Philippines, Indonesia, Malaysia, and Thailand, are higher than in any other region. An engagement system that cannot respond in real time to individual behavioural signals is not merely suboptimal in this context — it is invisible.
Regulatory fragmentation demands architectural flexibility. APAC is not one market. It is a collection of distinct regulatory environments — MAS TRM in Singapore, OJK in Indonesia, PDPA in Thailand, NPC in the Philippines, APPs in Australia — each with specific requirements around data residency, customer consent, and the deployment of AI-driven systems. An intelligence layer that cannot adapt its architecture to meet these requirements without rebuilding from scratch is not deployable across the region. This is not a compliance problem. It is a design requirement.
The competitive pressure is asymmetric. Tier-1 banks in APAC are not just competing against each other. They are competing against digital-first challengers that have no legacy infrastructure, no batch processing constraints, and no organisational inertia to manage. Challenger banks in the Philippines and Indonesia are deploying behavioural engagement mechanics that drive daily active usage across millions of customers. The incumbents with the largest customer bases have the most to gain from deploying an intelligence layer — and the most to lose from continuing without one.
The Window for First-Mover Advantage
53% of financial services firms are already deploying AI agents in production (ResearchAndMarkets, 2025). Almost all of those deployments are in risk, fraud, and back-office functions. The engagement intelligence layer is, at this moment, a white space — and white spaces in enterprise technology close faster than most institutions plan for.
The institutions that deploy a genuine revenue intelligence layer in the next 12 to 18 months will build an advantage that compounds. Better data produces better predictions. Better predictions produce more relevant engagement. More relevant engagement produces stronger behavioural signals and more data. The cycle is self-reinforcing — and it means that the gap between institutions that start now and institutions that wait is not linear. It grows.
The $60B AI investment in BFSI is real. The question is not whether AI will transform financial services — that transformation is already underway. The question is which layer of the business captures the next wave, and which institutions are positioned to benefit from it.
The back office has had its transformation. The risk function is in the middle of one. The customer-facing engagement layer is next. The institutions that recognise this now — and build the intelligence infrastructure to support it — will not just capture a share of the $60B. They will define what BFSI customer relationships look like for the next decade.
The back office has had its AI transformation.
The customer engagement layer is next. The institutions that move first will not just capture a share of the market — they will set the standard for what comes after.
The Question BFSI Leaders Should Be Asking Now
BFSI institutions are not short of AI ambition. The $60B market projection reflects genuine commitment to intelligent, data-driven operations across the industry. The gap is not in intent — it is in allocation.
The operational layers of financial services — risk, fraud, compliance, underwriting — have been rebuilt around intelligence. The customer-facing engagement layer has been left running on architecture that pre-dates AI as an operational concept. The consequences of that asymmetry are showing up in engagement metrics, retention rates, and the inability of most institutions to answer a simple CFO question: what did our customer engagement programme generate for the business last quarter?
Closing that gap does not require waiting for the next technology cycle. The data assets are already there. The AI capabilities are mature and deployable. What is needed is an architectural decision — to build the engagement layer that connects customer intelligence to revenue outcomes with the same rigour that has already been applied to every other layer of the BFSI business.
If your institution is working through what that layer looks like in practice, the Tier 1 Engagement Audit offers a structured starting point — a diagnostic framework for assessing where your current engagement stack sits against the standard that this market is converging on.
The back office has had its AI transformation.
The customer engagement layer is next. The institutions that move first will not just capture a share of the market — they will set the standard for what comes after.
If you’re mapping your current loyalty architecture against these use cases and want to work through what your signal coverage actually looks like, we’re happy to think through it with you.
FAQs:
How is AI being used in BFSI customer engagement?
AI in BFSI has been deployed primarily in fraud detection, credit underwriting, and risk management. Its application to customer engagement — predicting individual behaviour, recommending relevant interventions, and attributing engagement activity to revenue outcomes — remains the least developed application area. This is the next frontier for AI investment in financial services, and the institutions that build this capability first will define the category.
What is a revenue intelligence layer in banking?
A revenue intelligence layer connects customer data, behavioural signals, and business growth objectives into a single operating system for customer engagement. It unifies data, builds individual intelligence profiles, recommends growth strategies, executes them without IT bottlenecks, delivers personalised experiences, and attributes every engagement action to revenue outcomes. It is the system between knowing a customer exists and knowing how to grow the relationship.
Why is the AI opportunity in BFSI customer engagement underinvested?
Because the ROI case for engagement AI is harder to make than the ROI case for fraud AI. Fraud detection AI delivers measurable outcomes in weeks. Engagement AI requires measurement infrastructure — the ability to connect a behavioural nudge to an incremental revenue outcome — that most BFSI institutions have not yet built. The value is large; the measurement problem has delayed the investment.
What is the difference between personalisation and customer intelligence in banking?
Personalisation is a campaign capability — it delivers a tailored message to a defined segment. Customer intelligence is an operational capability — it builds an individual-level behavioural profile, predicts what a specific customer will do next, and recommends what the institution should do in response. Most BFSI institutions have personalisation. Very few have customer intelligence. The difference shows up in engagement rates, retention, and cross-sell velocity.
Why does BFSI in APAC need an AI engagement layer more urgently than other regions?
Three reasons: mobile banking penetration exceeds 90% in many APAC markets, raising customer expectations beyond Western benchmarks. Regulatory environments vary significantly across the region, requiring architecturally flexible deployment. And incumbent banks are competing directly against digital-first challengers with no legacy constraints — making engagement velocity and relevance a survival issue, not a competitive advantage.
What metrics should BFSI use to measure customer engagement intelligence ROI?
Revenue-connected metrics: incremental transaction volume attributable to engagement campaigns, churn defensibility score, cross-sell velocity from enrolment to second product adoption, cost per activated user, and incremental revenue per campaign. These replace proxy metrics like redemption rate, NPS movement, and campaign open rate — which measure activity, not commercial outcomes, and cannot survive CFO scrutiny.
Praveen Vadla
Praveen Vadla is Senior Digital Marketing Manager at Perx Technologies. With over 10 years of experience in B2B SaaS marketing across the US and Southeast Asia, he focuses on customer loyalty, engagement, and retention strategy. He writes on how brands build lasting customer relationships in a mobile-first economy. Connect with Praveen on LinkedIn.
Recommended for you
An ISO/IEC27001:2013 and ISO 27018:2019 certified cloud solution
SAY HELLO
Perx Technologies Pte Ltd
20A Tanjong Pagar Road
Singapore 088443
An ISO/IEC27001:2013 and ISO 27018:2019 compliant cloud solution
An ISO/IEC27001:2013 and ISO 27018:2019 certified cloud solution