• A major Southeast Asian telecom provider acquired 30% of its postpaid users from prepaid conversions
• However, the segment showed only a 5% conversion rate and a 30% default rate, resulting in over $6.5M in annual revenue leakage
• The goal was to replace basic rule-based lead filtering with dynamic models using richer data

• Developed a framework to select leads based on each customer’s incremental value
• Integrated Telco data (call records, billing, CRM, top-up) with external sources
• Inferred user mobility patterns via cell tower proximity to home, work, and frequented areas

• Captured 60% of likely defaulters, saving ~$4M annually‍
• Reduced lead volume 10x (lowering costs) while improving conversion by 2%, driving ~0.3% top-line growth

• Negligible MAU growth, as acquisition matched churn
• Challenged to achieve high MAU and fast growth targets
• Goal: Identify triggers to drive user growth and app engagement

• Rapid experiments to compare active vs. inactive user traits.
• Used behavioral data across categories such as:
  
  • Content consumption, user events, traffic source, device attributes
  • DWH, CDR, and browsing data

Insights led to initiatives that resulted in ~7% reduction in leakage, including:

• Optimized Earn and Burn strategy to boost stickiness
• Improvements to UI, chat, and interaction functionalities
• General app performance enhancements and bug fixes

• One of Thailand’s largest telecos supporting retail operations faced location strategy uncertainty due to generational and COVID-driven behavioral shifts‍
• The goal was to find strategic sites for new stores and prioritize existing ones based on performance

• Connected telco footfall, segmentation, retailer data, and OSM to uncover location performance drivers
• Analyzed population foot traffic and linked segments to high-performing stores
• Forecasted sales at potential new sites using explainable factors for decision support

• Built highly accurate predictive models for store performance
• Delivered explainable insights into location characteristics based on evolving customer behaviour
• Enabled smarter prioritization and data-driven expansion planning

• One of Thailand’s largest Telco operators with over 30 million customers
• Challenged to increase SMS campaign performance in terms of conversion rate and total sales
• The goal was to identify the right target audience with the most effective message and timing

• Connect the dots across multiple datasets including profiles, call detail records, payment history, loyalty programs, usage patterns (voice, SMS, data), browsing behaviors, and location data
• Tested hundreds of millions of hypotheses to identify features linked to conversion and high-value customers
• Extracted target audiences through micro-segmentation and propensity models

• Boosted campaign response rate by 270%, driving customer satisfaction and increasing sales by 5.5M THB‍
• Improved customer understanding using explainable features and ongoing profiling

• Leading North American pay TV and broadband provider
• Suffering from extremely high voluntary churn across products
• Existing models failed to identify enough high-risk customers
• Goal: Improve retention by accurately identifying churn-prone users and adapt over time

• Connected profile, payment, loyalty, usage, and browsing data across services, including viewing history and location
• Continuously tested millions of hypotheses to uncover churn drivers
• Extracted high-risk microsegments as precise campaign targets

• Identified churn segments with 5–10x higher propensity than the base rate
• Pinpointed 20% of churners with actionable reasons, enabling tailored interventions
• Achieved $100M+ annual run-rate impact through improved targeting