Retail FX — Activation, Reactivation & Deepening
A repeatable analytics-led growth engine for activation, reactivation and deepening across a 2 Mn customer universe.
- 2 Mn customer universe across two entities
- Top 45K high-potential customers prioritized every 3 months
- Digital-only execution: email, SMS, push & in-app RTIM
From business problem to measurable impact.
Business Problem
A large retail customer base had low or uneven FX engagement: many customers were inactive, under-engaged or conducting foreign-exchange activity outside the bank. Uniform mass communication could not distinguish latent potential from low-propensity customers.
My Role
Led the end-to-end analytics-to-campaign design: translating the FX growth problem into propensity use cases, shaping the feature universe and model-to-campaign logic, defining digital channel execution and linking campaign economics to conversion and incremental income.
Solution
Machine-learning propensity models rank high-potential customers for activation, reactivation and deepening. The top cohorts are refreshed on a recurring cycle and engaged directly through email, SMS, push notifications and contextual RTIM placements within app journeys — with no RM hand-off.
Technology & Methods
XGBoost · decision-tree / random-forest challengers · 300+ engineered features · 20+ internal systems · CASA & balance behaviour · cross-border / exchange-house signals · card behaviour · digital engagement · model monitoring
Leadership Lesson
“For high-volume retail banking, the execution layer can be fully digital: model scores create value only when they are connected to relevant journeys, offer economics and a closed measurement loop.”
Detailed Case Study
The infographic above is the executive summary. The full portfolio document contains the methodology, operating model, governance and implementation narrative at an externally shareable level.
View detailed PDF ↗