In a previous analysis, we diagnosed the structural challenge of Paycheck Forwarding: banks receive the monthly deposit, but daily spending flows to competitors. The bank carries the infrastructure. The competitor owns the relationship. For banks in this position, Primary Financial Institution (PFI) status erodes with every paycheck.
Most retail executives recognize this. The challenge is how to fix it without replacing your core or hiring an army of data scientists.
Leading banks are answering that question by moving from static servicing to growth orchestration: a progressive, risk-adjusted framework that layers intelligence on top of existing infrastructure and turns deposit relationships into primary ones.
Below we share what that operating model looks like in practice.
The operational shift: agentic AI as a force multiplier
The biggest barrier to reclaiming PFI status is operational bandwidth. Your teams cannot manually analyze transaction data, segment audiences, and craft thousands of unique retention offers for millions of customers. The math doesn't work.
Banks are deploying Agentic AI to help bankers in layers to:
- Monitor transaction history continuously to spot churn signals like external transfers
- Propose engagement strategies based on behavioral patterns
- Generate personalized retention content at the individual customer level
Critically, this operates on a Human-in-the-Loop principle. The AI proposes, and your team approves. You retain strategic control while removing the manual workload.
But the operational model is only as good as the intelligence feeding it. What matters is whether your AI detects the right moments to act on. This is where the intelligence layer comes in.
The intelligence layer: detecting meaningful events
Demographics don't predict intent. Knowing a customer's age, income, and zip code tells you almost nothing about where they will move their money next.
Banks reclaiming PFI status are tracking meaningful events in the form of behavioral triggers that reveal intent in real time. We share three examples:
The cash leak (deposit retention)
The signal: A customer receives their paycheck. Within 48 hours, $1,000 moves to an external brokerage or high-yield savings account.
The fix: Catch this pattern and offer a competitive savings product before the next transfer. This turns deposit flight into deposit retention.
The freed-up cash flow (share of wallet)
The signal: A customer makes their final car loan payment. They now have $500/month in disposable income.
The fix: If you catch it before they spend it, you can propose an investment account or credit product that deepens the relationship.
The mortgage shopper (anchor product)
The signal: A customer pays an appraisal fee or searches "current rates" in your app.
The fix: This is the highest-value churn signal in banking. Detect it early, and you are competing for the mortgage - not processing the exit.
The pattern across all three: detect the moment, propose the right product, deepen the relationship. Each interaction moves a customer from single-product user toward the PFI threshold.
The implementation roadmap: pilot, prove, scale
The fear of a "Big Bang" implementation paralyzes decision-making. Luckily, the most effective PFI recovery programs avoid it entirely. They follow a progressive path that delivers value in stages and does not require replacing your core.
Growth orchestration functions as a layer on top of existing systems. It ingests data from the core, enriches it with intelligence, and triggers actions in the engagement layer: the app, the branch, and the contact center.
Phase 1: the pilot (months 1-6)
Focus on your top 10% of customers (Mass Affluent). Connect an orchestration layer to existing transaction data. Identify one meaningful event - start with cash leaks - and run a single orchestration play.
The goal: validate that timely intervention reduces outflow and increases product holdings.
Phase 2: expansion (months 7-12)
Move to the Emerging Affluent segment. Expand AI agents to identify cross-sell opportunities based on propensity scoring and meaningful event detection.
The goal: move customers from single-product users to multi-product relationships - the foundation of PFI status.
Phase 3: scale (months 13+)
Deploy self-learning models across the full customer base, optimizing engagement timing, channel, and content for millions of users.
The goal is lifecycle automation across all segments.
Proof it works
Bank Muamalat: from legacy infrastructure to competitive deposit capture
Malaysia's largest Islamic bank faced a familiar challenge: fragmented legacy systems that couldn't support personalized, real-time engagement. Rather than attempting a full core replacement, Bank Muamalat took a layered approach - deploying an AI-powered engagement platform on top of its existing infrastructure to launch ATLAS, Malaysia's first Islamic digital-only bank.
The results, within three months of launch: 58% of new users were new-to-bank customers, and 96% of deposits came from fresh funds transferred from other banks. Not rebalanced internal accounts, but fresh competitive wallet share, captured through a digital experience that didn't exist six months prior.
Techcombank: scaling digital engagement across 13.4 million customers
Vietnam's fifth-largest private bank had outdated systems that couldn't scale engagement across a growing customer base. Techcombank deployed a unified engagement platform on top of its existing core - starting with retail, then reusing proven journeys to launch business banking in just six months.
Over 50% of Techcombank's total savings and investments now originate from digital banking - direct evidence of deposit relationships deepening through the digital channel. They acquired 1.9 million new customers digitally in 2024 and achieved 51% growth in retail e-banking transaction volume. Their app now holds a near-perfect 4.9 rating.
The path forward
Three out of four of your customers bank somewhere else too. With margins compressing, the cost of inaction isn't just missed revenue - it is structural decline.
The barrier to act is lower than most retail leaders assume. Growth orchestration layers on top of existing systems - no core replacement required. You can modernize your engagement model today, even if your core modernization is years away.'
Watch: Turn single-product customers into lifelong relationships
See how banks use AI agents to scale personalization without losing control. This session breaks down the mechanics of the Human-in-the-Loop architecture in practice.
