BankAI Churn Shield
AI & Analytics

BankAI Churn Shield

Nusabank Digital
Financial Services
18 weeks
Pythonscikit-learnSalesforce Einstein

Project Overview

Built an AI-powered churn prediction engine processing 2M+ customer signals daily for a regional digital bank, with automated intervention triggers.

The Challenge

Nusabank Digital was experiencing accelerating customer churn in a highly competitive market. Their reactive retention team was contacting customers only after cancellation requests — too late to intervene. They had no early-warning system and lacked confidence in their data quality.

Our Solution

LOYA built an XGBoost churn prediction model trained on 18 months of transaction, engagement, and support data. The model runs daily on 2M+ signals and automatically enrolls at-risk segments into personalised retention journeys in Salesforce Marketing Cloud, with call-centre escalation triggers.

Project Gallery

XGBoost churn model — feature importance and prediction confidence scores
Real-time churn risk dashboard — at-risk segment monitoring
Retention journey workflow — automated intervention triggers in Salesforce Marketing Cloud
Model validation results — AUC curve and precision-recall metrics

Project Videos

BankAI Churn Shield — AI Model Architecture & Results Walkthrough

Downloadable Documents

Related Projects

TelkomCRM Migration
CRM Implementation

TelkomCRM Migration

Migrated a leading telco from a legacy CRM to Salesforce with custom AI lead-scoring modules, 360° customer view, and automated renewal workflows.

RetailConnect Automation
Marketing Automation

RetailConnect Automation

Implemented full lifecycle marketing automation for a national retail chain — from acquisition nurture sequences to win-back and loyalty campaigns.

CDP Nusantara
Customer Data Platform

CDP Nusantara

Unified 7 disparate data sources into a single real-time customer profile platform for a fast-growing e-commerce conglomerate across Indonesia.