An ISO/IEC27001:2013 and ISO 27018:2019 certified cloud solution
80 Key Customer Loyalty Stats & Trends to Watch in 2026 | Perx
Blogs
From Understanding Agentic Loyalty to Deploying It: A Sequencing Guide for BFSI Leaders
Capgemini says 75% of banks plan to adopt AI agents in customer service within two to three years. Far fewer have deployed. That gap between intent and execution is where the next round of competitive advantage in banking loyalty will be won or lost.
IN BRIEF
- Capgemini’s Banking Top Trends 2026 names AI-driven loyalty, agentic AI, and gamification as the three plays banks are using to win customer engagement. The direction is now an analyst consensus, not a vendor claim.
- The market signal is strong but the deployment is not: 75% of banks plan to adopt AI agents in customer service, yet most loyalty programmes still run on human-designed, human-triggered campaign cycles.
- Moving to agentic loyalty is a sequencing problem, not a switch. Teams that try to jump straight to autonomous decision-making without the data and guardrail foundations underneath it tend to stall in pilot.
- A verified APAC banking deployment shows what behaviour-triggered loyalty delivers at scale before full autonomy even enters the picture: 32x ROI and 709,000 users activated with Jenius (SMBC Indonesia).
If you have read our guide to what agentic AI means in loyalty programmes, you already understand the architecture: the four-layer Agentic Loyalty Stack, the six autonomous revenue agents that sit at the top of it, and why a monthly campaign cadence can no longer keep pace with individual customer behaviour.
So this is not another explainer. The question that matters for a senior BFSI team is no longer what agentic loyalty is. It is how to get from where the programme runs today to where the analysts say it needs to be, in an order that survives a risk committee, a CFO, and a two-year IT roadmap. This piece is about sequencing.
The direction is no longer in dispute
For most of the last decade, the case for AI-led loyalty in banking was a vendor argument. That has changed. In its Banking Top Trends 2026 report, the Capgemini Research Institute for Financial Services names three plays banks are using to connect with customers, and all three sit inside the loyalty and engagement layer: implementing agentic AI in customer contact centres, using gamified mobile-first platforms, and leveraging AI-powered loyalty programmes to personalise offers and services.
Capgemini frames AI-driven personalisation as a growth driver in its own right, arguing that financial institutions should shift to differentiated loyalty programmes that personalise rewards rather than offering generic incentives on data they already hold. The supporting evidence is worth putting in front of a board.
Capgemini Banking Top Trends 2026 – the case for AI-driven loyalty
| Signal | What it tells a BFSI leader |
|---|---|
| 38% | Of customers who switched financial institutions in 2024 did so because they were not satisfied with service quality. The churn driver is experience, not price. |
| 73% | Of card customers are motivated by personalised offers and rewards, which makes generic incentives a measurable revenue leak. |
| 75% | Of banks plan to adopt AI agents in customer service functions within the next two to three years. |
Source: Capgemini Research Institute for Financial Services, Banking Top Trends 2026. Figures drawn from the published report; confirm against the source before citing externally.
Read the third number again, because it is the one that should shape your planning. Three-quarters of banks plan to adopt AI agents. Planning is not deploying. The market has reached consensus on the direction and has not yet moved on the execution. That is precisely the position in which first-mover advantage is still available.
The consensus has formed around the destination. It has not formed around who arrives first.
Why most agentic loyalty efforts stall in pilot
The failure pattern is rarely a failure of ambition. It is a failure of order. Teams read that agentic AI can detect a spend decline on Tuesday and intervene on Wednesday, and they try to buy the Wednesday intervention without first building the thing that detects Tuesday. Autonomous decisioning sits at Layers 3 and 4 of the stack. It cannot run on a foundation that does not yet exist at Layers 1 and 2.
There are three foundations that have to be in place before an autonomous agent can act safely inside a regulated bank:
Signal density before autonomy
A retail loyalty programme might see a customer a few times a week. A bank sees them dozens of times a day across channels. That signal density is what makes an agentic intervention precise rather than generic, but only if the signals are unified. Fragmented data is the single most common reason an agent has nothing useful to act on.
Guardrails before action
An autonomous agent in BFSI does not get to override compliance. It optimises within defined limits, on incentive disclosure, communication frequency, and data usage. Those limits have to be designed and agreed before the agent is allowed to execute, not retrofitted after a pilot raises a flag.
A defined behavioural target before a reward
An agent with a vague objective produces vague outcomes. The objective has to be specific and measurable, reduce 90-day primary-account dormancy, lift credit-card reactivation, before the system has anything to optimise toward. A deployment without a measurable behavioural target is not an agentic deployment. It is automation with better marketing.
A sequencing model that survives a risk review
The progression below is deliberately ordered so that each phase produces a defensible result before the next one starts. The point is not speed for its own sake. It is to reach autonomous decisioning with the data, the guardrails, and the internal trust already in place.
PHASE 01 · FOUNDATION
Unify the behavioural signal
Bring the high-frequency signals a bank already generates into one behavioural layer. The deliverable is not a campaign. It is a single, real-time view of customer behaviour that later phases can act on. Most BFSI organisations sit here today.
PHASE 02 · PREDICTION
Add predictive intelligence and prove it quietly
Layer predictive models onto the unified signal, spend-decline detection, lapse probability, redemption likelihood, and validate their accuracy against real outcomes before anything acts on them. This phase builds the internal evidence the risk committee will ask for in Phase 3.
PHASE 03 · GUARDED AUTONOMY
Let one agent act inside tight guardrails
Move a single, well-bounded use case, often offer optimisation, from prediction to autonomous execution, within strict compliance limits and on a defined behavioural target. One agent, fully measured, builds the trust that funds the rest.
PHASE 04 · SCALE
Extend to the full agent set
With the foundation, the evidence, and the guardrails proven, extend autonomy across the remaining revenue agents. This is the Layer 3 to Layer 4 progression Capgemini’s two-to-three-year window is describing, reached on a footing that holds up under scrutiny.
PROOF POINT · APAC BANKING
What behaviour-triggered loyalty delivers before full autonomy
It is worth being precise about what is already proven versus what is still forward-looking. The deployment with Jenius (SMBC Indonesia) was behaviour-triggered, rules-based loyalty, behaviourally designed and measurable, not fully agentic AI. It is a marker of what the foundation phases can deliver at scale, and a sense of the headroom the agentic layers add on top.
32x
Return on investment
709K
Users activated
What this means for your next planning cycle
The honest read of the Capgemini data is that the industry has agreed on where loyalty is going and has not yet moved. For a BFSI team, that produces a narrow and valuable window. The work that creates advantage in the next twelve months is not buying an autonomous agent. It is building the foundation that an autonomous agent needs, in an order that earns internal trust at each step, so that when the rest of the market begins deploying in two to three years, your programme is already there.
That is the difference between a platform that drives revenue and one that merely reports on it. The destination is settled. Sequencing is the advantage.
FAQs:
What is the difference between agentic AI and the AI our loyalty platform already uses?
Most loyalty AI today is predictive or assistive: it tells a team that a customer is likely to churn, then waits for a human to decide and act. Agentic AI acts on the prediction autonomously, selecting an intervention from a defined toolkit, executing it through the right channel at the right time, and recording the outcome to improve future decisions, all within compliance guardrails. The defining difference is autonomous execution against a measurable objective.
Can agentic AI operate inside BFSI compliance requirements?
Yes, when it is designed to. A well-built agent does not override compliance rules on incentive disclosure, communication frequency, or data usage. It optimises within those defined limits. The guardrails have to be agreed before the agent is allowed to execute, which is why guarded autonomy sits at Phase 3 of a sound sequencing model rather than at the start.
Why not move straight to autonomous decisioning?
Because autonomous agents sit at the top of the stack and depend on foundations beneath them: unified behavioural signal, validated predictive models, and agreed guardrails. Skipping to execution without those in place is the most common reason agentic loyalty efforts stall in pilot. Sequencing exists to reach autonomy on a footing that survives a risk review.
How long before the rest of the market deploys?
Capgemini’s Banking Top Trends 2026 reports that 75% of banks plan to adopt AI agents in customer service functions within two to three years. That timeline is the planning window. Teams that build the foundation now are positioned to be live when the broader market is still beginning.
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