Banco Caja Social
Overview
Section titled “Overview”| Field | Detail |
|---|---|
| Industry | Banking |
| Country | Colombia |
| Channel | Outbound voice + later inbound voice |
| Initial delivery | 19 days for collections agent |
| Primary use case | Debt collection and subsequent weekend customer service |
Business challenge
Section titled “Business challenge”Collection calls were labor-intensive with a 45% contact-to-promise baseline; incoming weekend service demand lacked sufficient scalable coverage.
AI solution
Section titled “AI solution”“María” calls customers in early delinquency, retrieves account details, discusses payment commitments, and records outcomes in the collections platform. A second agent, “Gloria,” handles inbound service questions across five domains.
Technical implementation & delivery
Section titled “Technical implementation & delivery”Co-developed account/collections API, real-time result write-back, speech turn-taking and pace adaptations, governance checks; Gloria also uses a 190-document knowledge base and Genesys Cloud telephony.
Results & KPIs
Section titled “Results & KPIs”| Metric | Reported result | Measurement scope / context |
|---|---|---|
| Promise-to-pay rate | 65% vs. 45% human baseline | Among direct contacts; not an all-dial conversion |
| Average handling time | 33% lower | Collections agent |
| Payment conversion after promise | 37% vs. 31% | Share that actually paid, as reported |
| Early call volume | 43,658 calls in 3 weeks | Collections agent María |
| Weekend inbound coverage | 38% of weekend calls | Later agent Gloria |
Key takeaway
Section titled “Key takeaway”This is the strongest revenue-linked KPI case: payment promises, actual payments, throughput, and cost per call.