How ICBC’s Software Center Built a Penetrative Risk Management System for Credit‑Card Partners

Regulatory background and partner‑institution challenges

The Central Financial Work Conference mandated comprehensive strengthening of financial risk prevention, emphasizing "penetrative management" of partner institutions. ICBC’s credit‑card center, overseeing 28,000 partners, faced three core problems: delayed risk identification relying on static documents, insufficient responsibility penetration across layered ownership, and weak, fragmented risk‑control fuse mechanisms.

System architecture and core capabilities

ICBC’s Software Development Center created a "penetrative" management system driven by a "full‑lifecycle management platform" and a "data‑intelligence hub." The platform delivers three capabilities: intelligent admission screening, dynamic risk‑signal fuse, and precise ecosystem empowerment, forming a pre‑, during‑, and post‑risk control loop.

Dynamic risk perception breakthrough

By aggregating real‑time external data sources—business registries, judicial records, public sentiment, and operational metrics—the system builds a comprehensive, continuously updated digital portrait of each partner. An intelligent rule engine automatically captures 21 key risk signals (e.g., equity shocks, litigation spikes, abnormal operations) within seconds, shifting from monthly manual reviews to minute‑level real‑time alerts.

Deep association penetration

Leveraging the "RongAn e‑Xin" risk‑big‑data service, the platform identifies four association dimensions:

Equity penetration: traverses multi‑layer equity structures to reveal ultimate controllers.

Legal/personnel overlap: detects cross‑appointments of legal representatives, directors, and executives.

Physical/operational links: matches registered addresses, office locations, and phone numbers to spot concentration or reuse.

Equity‑chain decomposition: reconstructs complex equity, investment, and guarantee chains, producing a full association network.

This deep analysis uncovers hidden institution clusters that could enable joint arbitrage, fictitious transactions, or regulatory evasion, thereby reinforcing the systemic‑risk firewall.

Automated risk‑control system

The solution connects risk‑signal detection to business‑system APIs, enabling automatic strategy execution. A dynamic risk state machine with elastic thresholds reduces transaction limits by 50% when an institution’s rating jumps to high‑risk and restores thresholds over time. The "ICBC TianCe" intelligent core refreshes risk portraits every minute, compressing the entire risk‑handling workflow to two minutes and achieving 100% blockage of fraud groups while preserving zero‑impact operation for compliant partners.

Future direction

Looking ahead, the team plans to embed artificial‑intelligence models for finer‑grained risk scoring, implement dynamic tiered rating schemes, and apply privacy‑computing techniques to safely match anonymized customer profiles, further enhancing ecosystem collaboration and marketing conversion for partner institutions.