Author: Jingdong Insurance Xue Yankai
I. Introduction
When it comes to JD.com, most people's first impression is that it is a bustling shopping mall, and the convenient shopping experience is deeply rooted in the hearts of the people. But in this fertile land of commerce, there is also a group of special JD people - they do not sell "goods", but take it as their responsibility to deliver protection and protect happiness, that is, JD insurers. My department is the R&D department of insurance agency business, focusing on providing a full range of technical and business support for insurance agents, and we are committed to continuously optimizing services, creating more opportunities for agents, and jointly promoting the prosperity and progress of the insurance industry.
Agents are an indispensable bridge of trust between customers and the company, and their role is particularly important in the ever-changing and challenging business environment. In order to maintain the stability of this bridge of trust and ensure the orderly and efficient operation of the business, the establishment of rules and regulations has become a crucial task. It is based on this demand that the Basic Law came into being, which is not only a set of management and salary systems, but also the crystallization of wisdom to guide and regulate the operation of the industry. The Basic Law is the cornerstone of the healthy development of the insurance agency business, ensuring that each agent can reach his or her potential within a clear framework and receive generous commission rewards.
In order to improve the service efficiency of agents and back office, it is necessary to build an intelligent, automated, and personalized system architecture to optimize and innovate business processes, and at the same time, it can quickly respond to market changes and accurately assess risks. However, the existing architecture no longer meets the needs of the current environment, and the system issues are frequent, which not only puts a heavy burden on the business team, but also seriously affects the company's operational efficiency and customer trust. In the face of this problem, we are deeply aware of the urgency and necessity of refactoring the system. Next, let's walk into the road of transformation of the Basic Law system.
II. Introduction to the Basic Law
The Basic Law on Insurance is a basic regulation for the day-to-day management of insurance marketers (including agents and practitioners), rather than a uniform legal provision. These regulations cover business development, promotion, rewards and punishments, and profit distribution, and are the core tools for insurance companies to promote the development of their marketing teams. In JD's insurance agency business, the Basic Law of Insurance mainly serves the company's field business personnel - agents, and guides the career development, performance appraisal and salary distribution of agents by setting clear rules and standards. This is not only related to the income and development of agents, but also directly affects the stability and market competitiveness of the insurance agency business.
2.1 Core content of the Basic Law
Daily management: including the daily behavior norms such as signing, changing, and terminating the contract of the agent, and providing staff training to ensure the stability and professionalism of the team.
Compensation and benefits: The calculation method of remuneration is clarified, including direct commission, indirect commission, non-cash incentives, etc., and employee benefits, such as pensions, etc.
Performance appraisal: Agents from junior salesmen to business area directors, each stage has clear performance requirements and corresponding incentives, and sets a clear promotion path. Evaluate from multiple dimensions such as business achievement rate, continuation rate, and team indicators to ensure the quality of the agent's work.
2.2 System service link process
III. Challenges to the Basic Law
3.1 System Complexity
Data processing complexity: As the scale of the business expands, a large amount of business data needs to be processed, including customer information, policy information, performance data, etc. Due to the diversity and complexity of data, system construction is facing great challenges.
System customization requirements: There are three sets of versions of the Basic Law, involving a total of 36 commission items and 17 ranks, and there will be new requirements for the Basic Law in the future. The three sets of Basic Laws are quite different from each other, which requires the system development team not only to have deep knowledge of the insurance business, but also to have strong system customization capabilities to meet the specific needs of the business team for different versions of the Basic Law.
Changes in policies and regulations: Policies and regulations in the insurance industry change frequently and are subject to strict regulation. In order to adapt to policy changes, update the corresponding Basic Law rules in a timely manner, and ensure compliance with the Basic Law, the system must be flexible and scalable.
3.2 Business Complexity
Diversification of product categories: Insurance companies offer a wide range of products, including life insurance, property insurance, car insurance, etc. Each product has its own unique insurance terms and rate structure, which places higher demands on the formulation and enforcement of the Basic Law.
Diversification of sales channels: The insurance agency business sells products through a variety of channels, including agent offline channels, online sales channels, etc. Different channels have different sales models and customer needs, which requires the Basic Law to be flexible in responding to various sales scenarios.
Risk management: Insurance business itself has a high degree of risk, including credit risk, market risk, operational risk, etc. The Basic Law needs to strengthen risk management and ensure the sound operation of the company while promoting business development.
Fourth, the transformation of the Basic Law system
4.1 System full-scenario architecture
4.2 Comparison of system reconstruction
| Before refactoring | After refactoring | |
| readability | The code is highly coupled, repetitive, confusing, and difficult to understand and modify | The code is low coupling, eliminates duplication, and has clear logic Reduced complexity and easy to understand |
| System flexibility | The low modularity of the system is not conducive to maintenance and expansion | Modular design, easy to maintain and expand |
| reliability | 0% | 99.99% |
| security | Error handling mechanism is missing | Automatic review and link tracing |
| performance | Relying on real-time data, there is unnecessary resource consumption, long system response time, low throughput, and 6 hours of execution time | The metrics are preloaded and the drools rule engine is introduced, and the response time is short, and the execution time is 30 minutes |
5. In-depth introduction to the key breaking points of the system
5.1 Modular decoupling, clear division of responsibilities
Component-based design: Component-based design is the core strategy of modern software architecture, which subdivides a complex system into multiple independent and autonomous components, each of which focuses on a single business area (such as basic rules, performance rewards, position-level management, performance evaluation, team collaboration, etc.), which greatly improves the maintainability, scalability, and reusability of the system. These components encapsulate detailed business logic, granular data models, and standardized service interfaces to ensure low coupling and high cohesion between components.
Configurable: Gives the system a high degree of flexibility, allowing you to adjust the behavior and properties of components through an intuitive configuration interface, and quickly respond to changes in business needs without code modifications. This feature greatly simplifies the process of customizing basic law rules and reduces O&M costs.
Free combination: Users can flexibly select and combine versions of different components according to specific scenarios like building blocks, and quickly build a basic law that meets individual needs, which greatly shortens the implementation time of the basic law and meets the diversified needs of the market.
5.2 Introduce a rule engine
In view of the diversity of commission and assessment rules in the Basic Law, the drools rule engine is introduced as a solution. The declarative language DRL (DRL) greatly simplifies the process of describing rules, makes business logic more intuitive and easy to understand, and significantly improves the flexibility and efficiency of rule writing.
Decoupling of rules and code: By storing business rules in the form of scripts in files, rules and business code are decoupled, which is convenient for rule modification and management.
Extract and preload common indicators: Identify and extract common indicators in commission calculation and performance appraisal, and reduce the data access pressure of real-time calculation through the data preloading mechanism.
Rule Configuration:
Visual configuration interface: Provides a friendly page view to present extracted indicators (FYC, RYC, R13, etc.), conditional symbols (such as greater than, less than, equal to, etc.), algorithm symbols (such as addition, subtraction, multiplication, division, etc.), and logical symbols (such as or, and so on) to business or R&D personnel as options. Users can easily assemble the required business rules by selecting the method without having to go deep into the syntax details of the DRL script.
Automatically generate DRL scripts: After a user completes the rule configuration, the system automatically generates the corresponding DRL script based on the user's configuration options. This feature not only greatly reduces the coding workload of R&D personnel, but also ensures the accuracy and consistency of rule configuration, effectively avoiding rule execution problems caused by human error.
Intuitive reading of rules: Users can clearly see the composition and logical relationship of each rule through the rule configuration results displayed in the page view, and understand the meaning of the rule without reading complex DRL scripts. This intuitiveness not only reduces the learning cost for configurators, but also improves the efficiency of rule review and validation.
5.3 Multi-dimensional execution + data version isolation mechanism
The Basic Law can be disassembled into different dimensions of the Basic Law version and business teams to perform computing tasks, or the team, commission, and assessment components can be executed separately, which is more flexible and efficient than the previous architecture. At the same time, the calculation result data is isolated by data version, which can support the repeated execution of tasks and compare multiple versions of data, providing solid technical support for commission and assessment prediction. Commission forecasting allows agents to stay on top of task progress in real-time, including key information such as the number of orders completed, the commission revenue earned, and the gap between the current commission and the next commission tier. This information allows agents to recognize their room for improvement, stimulate the desire to challenge, improve their motivation to work, and strive to reach a higher commission level.
The implementation of this technology not only paves the way for the accurate prediction of commissions and assessments, but also gives unprecedented flexibility to the formulators of the Basic Law. In the process of formulating the Basic Law, it is easy to switch between different rules for combination, and the feasibility of the scheme can be verified in real time, which effectively reduces the cost of trial and error and shortens the decision-making cycle. For the implementation of the new Basic Law proposal, data-driven verification and optimization are carried out to ensure that the new proposal is not only in line with market trends, but also in line with the actual situation of the team.
5.4 Improve the R&D verification mechanism
5.4.1 Data Verification
To ensure that the commission accounting for this month's reconciliation fee is accurate, we have added a reconciliation process before the business review. Based on the preset proportion of handling fees, the system automatically summarizes direct commissions (direct commissions), indirect commissions (intermediate) and other types of commissions, and compares the summary results with the preset commission ranges to ensure that all commission expenses are strictly controlled within a reasonable range, effectively avoiding human error. It not only protects the due interests of the agent, but also maintains the financial health of the company. This optimization mechanism not only improves work efficiency, but also significantly enhances the accuracy and reliability of data processing.
5.4.2 Multi-dimensional business data analysis
It is connected to the BI reporting system to realize multi-dimensional visual analysis of data, opening up a new path for data verification and insight. Not only does it deepen the understanding of data complexity, but it also significantly enhances the interpretability and trustworthiness of data. Charts and dashboards can be easily built through a drag-and-drop interface, which can intuitively display the trend and correlation of data in different dimensions, so as to quickly identify outliers and further verify the accuracy of data. In addition, the multi-dimensional perspective also promotes cross-departmental data sharing and collaboration, jointly mining the business value behind the data, and providing solid data support for enterprise decision-making.
Sixth, the construction of system stability
6.1 Alarm mechanism
The system is connected to the application monitoring (UMP) and machine monitoring (MDC) monitoring systems, and is equipped with alarm items such as jvm monitoring, CPU usage, disk usage, connectivity, RSS memory usage, system load, TCP connections, network inflow rate, network outflow rate, and TCP retransmission number, so as to ensure that the application can run continuously, stably and efficiently when handling Basic Law tasks.
6.2 Basic Law task execution link tracking
Access link monitoring (PFinder): Relying on PFinder capabilities, it can monitor by multiple dimensional indicators, automatically bury the performance of common middleware such as SpringMVC, JSF, MySQL, and JMQ, automatically sort out the dependency topology of the upstream and downstream of the service and middleware, and quickly analyze the performance bottleneck based on the cross-service call tracking of the request when an exception occurs.
System task link monitoring: Key tasks (such as team tasks, commission settlement, performance appraisal, etc.) are executed according to the predetermined rules, the execution information of each node is retained, and the task is scanned through the scheduled task whether the execution of each node is completed, and the overtime is not executed for retry, and the email and telephone alarm will be triggered if the retry is not successful for 3 times, and the R&D personnel will respond to the alarm at the first time to ensure that the key task can be executed smoothly in accordance with the predetermined rules.
7. Summary outlook
Since the implementation of the system reconstruction, the system has undergone many rounds of rigorous task tests, and its accuracy rate has jumped to 99.99%, providing solid technical support and guarantee for more than 40,000 agents, completely reversing the old concept of the system for agents and the internal team, and has been affirmed by the business side.
In the future, we will unswervingly focus on the continuous improvement of system performance, stability and security to ensure that the optimization of each link can inject strong impetus into the rapid growth of the business." At the same time, we will actively embrace new technologies, continue to explore their application potential in the system, and strive to promote the system to a more intelligent and automated direction. We believe that through the continuous innovation and application of technology, we can further unleash productivity, accelerate the automation and intelligent transformation of business processes, and use technology as the engine to drive business development by leaps and bounds.