Client Service Optimization via Data

BNP Paribas Securities Services processes ~3 million Service Requests per year, exchanges 20-30 million emails, and operates across 5,000+ users on 1,300 desks via Hobart, their internal request management system. The system works, but at massive complexity — generating miscalibrated deadlines, heterogeneous workflows, invisible overloads, and unmeasured client friction.

As a Client Service Manager, how to improve daily client experience by better exploiting Hobart data to recalibrate SLAs, detect complex requests early, and reduce friction?

Service Request logs from Hobart (3M+ annual), email interaction data, desk performance metrics, SLA configuration data, request categorization (1,522 categories), and resolution time distributions.

1. 1.Recalibrated SLAs using P85 of observed resolution times (46% of deadlines were too generous, 44% too tight).
2. 2.Built a complexity detection system — 5% of SRs generate 15% of workload; flagged requests with >2 inbound or >5 interactions.
3. 3.Created a 'Strongest Desk' score (Speed × Volume × Reliability) for intelligent rerouting.
4. 4.Proposed clustering of 1,522 categories into 390 coherent clusters (4x complexity reduction).
5. 5.Designed an AI front-end: email → auto-classification → analyst validation → instant acknowledgment + Q&A suggestions.

Finished 2nd in the BDD challenge. Delivered 5 concrete solutions spanning SLA recalibration, complexity detection, intelligent routing, category simplification, and an AI-powered client interface.

Directly applicable framework for any large-scale service operations team dealing with high-volume request management and client satisfaction optimization.