BBVA drives 79% digital sales share with real-time AI-powered commercial product recommendation engine

BBVA, a bank operating across multiple countries, relied on expert-designed commercial campaigns built from limited datasets. A campaign targeting parents with an account for minors, for example, relied on indirect indicators such as school direct debits, an approach that was unscalable and failed to consider individual factors. These static campaigns could not adapt in real time or account for individual customer context, at a time when e-commerce was growing exponentially.

BBVA's internal AI Factory built a production-grade multiclass machine learning pipeline using LightGBM, a decision-tree-based algorithm, to predict which commercial product each customer is most likely to need at any given moment. Feature engineering converts raw behavioral and transactional data into meaningful signals (e.g., average supermarket spend, app activity patterns), and standardized global data sources, in the same format across all operating regions, allow the models to function across countries. CleanLab handles dataset imbalance by removing noisy observations from over-represented product classes, while Halving Search optimizes hyperparameters iteratively. The system integrates directly into BBVA's mobile app, delivering real-time, personalized offers at the individual customer level.

79% of BBVA's total sales are now completed through digital channels, with the recommendation engine operating across all of the bank's active regions. The shift from static, expert-designed campaigns to a continuously learning ML system produced measurable improvements in offer relevance and conversion precision. Key outcomes include:

• 79% digital sales share — the headline metric reflecting the scale of digital channel adoption supported by personalized recommendations
• Real-time personalization deployed at production scale across multiple countries from a single pipeline
• Continuous model improvement via online A/B testing, where new model versions are only promoted to production when improvements are statistically significant
• Fairness and explainability embedded into the evaluation cycle, reducing risk of biased recommendations across protected population groups

• Standardizing data sources across geographies allows ML models to function across a global institution's operating regions.
• Dataset imbalance is a material risk in product recommendation: tools like CleanLab should be applied proactively to prevent models from over-predicting high-volume products at the expense of lower-volume ones.
• Online A/B testing should gate model promotions — deploying only statistically significant improvements helps prevent regression.
• Fairness metrics (PPV, FPR, FNR) belong in the standard model evaluation checklist, not as an afterthought — especially when recommendations affect financial access across diverse customer segments.