Contact centre and collections transformation for a Mexican bank

The client is a Mexican commercial bank serving approximately 2.6 million retail and small-business customers through a national branch network, a mobile app and a contact centre. Its customer base includes a substantial cross-border segment who hold Mexican accounts and transact with the United States.

The bank's contact centre was failing at the most basic level. Call abandonment stood at 22.6%, meaning more than one in five callers hung up before reaching anyone. During month-end peaks it exceeded 33%. Customers responded by going to branches, which pushed urban branch queues past an hour and made the branch network the bank's most expensive service channel.

Collections was the larger problem. The internal team recovered 43% of balances in the one-to-thirty day bucket, against a benchmark the bank's own consultants placed near 64%. The team was small, worked business hours only and used manual dialling and spreadsheet tracking, so many accounts were never contacted before rolling into the next bucket.

Regulatory and operational reporting added a persistent drag, with back-office staff spending an estimated 900 person-hours monthly on manual compilation and reconciliation.

Building internally was not affordable. The bank's modelling showed that reaching the required capacity would strain its cost-to-income ratio, and that recruiting collections specialists at volume would take many months in the competitive Monterrey labour market.

The bank ran a formal tender with five bidders, three of them Mexican. Corpshore Mexico won on a combination of scale credibility and governance. The bank's risk committee was specifically concerned about outsourcing collections and customer data, and Corpshore's Toronto parent governance, combined with a documented LFPDPPP compliance position and the group's information security framework, satisfied a committee that had previously blocked outsourcing.

Corpshore also proposed taking collections and customer service together rather than separately, arguing that the two functions share the same customer and that splitting them would produce a fragmented experience. Monterrey's strong financial-services and business labour market gave the bank confidence in the talent.

The engagement launched with 44 agents and scaled to 146 across three functions: 84 in inbound service, 46 in collections, 16 in back office. Twelve team leaders, four QA analysts and two MIS analysts sit within the account. Service runs 07:00 to 22:00 seven days, collections 08:00 to 20:00 Monday to Saturday within permitted contact practices under Mexican consumer protection norms. All work is performed in Monterrey on the bank's core systems through a dedicated secure environment.

Abandonment before anything else. The first ninety days targeted a single metric. Capacity was sized against interval-level historical volume rather than daily averages, and a workforce management function was established where none had existed.

Collections segmentation. The prior approach treated all delinquent accounts identically. Corpshore introduced segmentation by balance, delinquency age, product and prior contact history, so high-recovery-probability accounts are contacted first and low-value accounts handled through lower-cost digital channels.

Contact strategy rather than dialling. Collections moved from manual dialling to a designed contact strategy with defined attempt patterns and channel sequencing across voice, SMS and WhatsApp, giving the bank a usable collections data asset for the first time.

Reporting automation. The back-office team automated the compilation and reconciliation steps in the reporting cycle, reducing a 900-hour monthly manual burden to a reviewed exception process.

Call abandonment fell from 22.6% to 3.4% by month 12. The bank's branch traffic data showed a 23% reduction in service-only branch visits over the same period.

Collections recovery in the one-to-thirty day bucket rose from 43% to 64%, moving the bank from below its benchmark to at it, with improvement across every bucket.

Reporting effort fell from approximately 900 person-hours monthly to 210, with the released capacity redeployed to analysis.