Galicia Seguros — predictive modeling and NLP across life, home, and auto.

A bank-owned insurer with millions of customers and rising acquisition costs.

Galicia Seguros operates life, home, and auto insurance lines across the existing customer base of one of Argentina's largest retail banks. The marketing organization had two compounding problems: outbound call campaigns were expensive and converting below target, and digital spend was being optimized on click metrics that didn't translate into policy purchases.

In parallel, the bank's own product team was marketing loans and cross-sell offers to customers without proper segmentation — sending the same offer to high-propensity buyers and unlikely converters alike. The cost of contact was constant; the yield wasn't.

Three model families. One operating principle.

Every contact with a customer is a cost. Every model exists to answer one question — is this contact worth making?

Boring stack, aggressive iteration.

No exotic tooling. The win came from disciplined experimentation, A/B holdouts measuring incremental lift, and tight collaboration with the marketing team.

Measurable lift in sales. Material reduction in call-center costs.

The work delivered double-digit improvements in conversion across campaigns and a substantial drop in cost-per-contact, measured against A/B holdouts retrained quarterly. Models remained in production with documented assumptions so the team could maintain them after the engagement.

Public-facing case studies for financial services clients are kept at this level intentionally. Specifics are shared in conversation, where context and constraints can be discussed properly.