SME Cross-Sell, Early Activation & Reactivation
ML-led targeting across international SME portfolios, connecting propensity scores to CRM execution and measurable income.
- 3 entities: Emirates Islamic, ENBD Egypt & ENBD KSA
- XGBoost propensity models + PySpark data preparation
- AED 40 Mn annual income; +12% for selected products
From business problem to measurable impact.
Business Problem
Recently acquired SME relationships had low product holding and low per-customer profitability. Product opportunities were spread across liabilities, FX and other banking needs, while sales capacity was limited and needed to be directed toward the customers most likely to activate.
My Role
Helped drive the full path from use-case design and PySpark-based data preparation through propensity modelling, CRM lead design, UAT, allocation/expiry rules, leadership cadence and measurement across three international entities.
Solution
Separate propensity signals rank customers for cross-sell and early activation, while a reactivation layer identifies abnormal declines in product income after accounting for trend, seasonality and one-off activity. Customer-level insights are surfaced in CRM to make the lead actionable rather than presenting a score alone.
Technology & Methods
PySpark · XGBoost · feature engineering · multi-dimensional banking data · transaction-network signals · CRM lead orchestration · UAT · monitoring and closed-loop measurement
Leadership Lesson
“A model is only one component of a growth engine. The harder leadership work is connecting data and modelling to CRM design, frontline adoption, lifecycle logic and measurable P&L.”
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 ↗