From Reactive Retention to Predictive Care: Re-engineering Customer Intelligence for a Multi-Million-Subscriber Telecom Base
Customer Intelligence & Propensity Modelling · Retention Strategy & Lifetime Value Management · Omnichannel Journey Redesign
Most organisations chase Lifetime Value the same way — more cross-sell, more campaigns, more outbound calls. Yet the moment a customer actually reaches out — to a call centre, an app, a store — that signal is almost never used. A customer who is one missed EMI away from suspension, or one bad network experience away from porting out, is too often handled by an advisor who has no idea any of that is happening. The result isn't just a missed save. It's a customer who told the business something was wrong, and was answered with silence.
This is precisely where a leading mobile telecom operator's postpaid business found itself.
Business Challenge
A large postpaid mobile business — a base of several million subscribers — was losing ground on every front at once:
- Monthly churn exceeding 2%, with nearly 5% of the base flagged as at-risk in any given month
- 75% of churn concentrated in two behaviours: port-outs to competitors (~45%) and prepaid downgrades (~30%) — customers who would rather keep their number than stay loyal to the brand
- Rising delinquency, with overdue payments cascading into involuntary barring and service suspension
- Weakening acquisition, as softening market perception (compounded by industry consolidation noise and broader regulatory overhang) dried up referrals
- An average loss of ~100,000 postpaid subscribers a month, each carrying meaningful monthly revenue value
Underneath the numbers was a structural problem familiar to most large consumer businesses: retention, collections, network, and customer care were each optimising for their own KPI, in their own silo, at the last mile — long after the customer had quietly disengaged.
Our Approach
Rather than treat this as a retention campaign problem, we treated it as a customer intelligence problem — one that needed a single, unified view of the subscriber before any engagement strategy could work.
Customer Intelligence — Listening to the Whole Customer, Not Just the Complaint
Propensity Modelling — Knowing What's About to Happen, Not What Already Happened
Retention Strategy & Lifetime Value Management — Acting Before the Moment of Churn
Journey Redesign — Removing the Friction Before It Becomes a Reason to Leave
AI Enablement Layer
Underpinning every step of this approach was a dedicated AI layer — the models, bots, and orchestration that made proactive, personalized intervention possible at scale.
- Churn propensity models
- Voice sentiment analytics
- Next Best Action engine
- WhatsApp Bot orchestration
- Intelligent customer segmentation
Business Impact
Within 9–12 months of deployment:
More importantly, the organisation moved from treating retention as a campaign to treating it as a capability — one where customer intelligence, ML-driven propensity, and journey design work together continuously, not in isolated sprints.
Why This Matters Beyond Telecom
The same pattern repeats across banking, insurance, hospitality, and retail: a customer's risk and opportunity signals exist somewhere in the organisation, scattered across functions that rarely talk to each other. The business that wins isn't the one with the most data — it's the one that can unify that data into a single customer view, act on it before the customer has to ask, and make every channel feel like it's the same relationship, not a different department.
This is the work we do at CX Pivot: building Customer Intelligence and propensity capability, designing Retention and Lifetime Value strategies that hold up under real operational pressure, and redesigning journeys so that personalization feels consistent — not coincidental — across every channel a customer chooses to use.