How to cross the financial data value gap from the comprehensive localization of the insurance database? Source

Translated

Data-driven is the basic form of digital transformation for enterprises. The most troubling thing for enterprises today is not the lack of data but the lack of data utilization; not the lack of understanding of the nature and value of data, but the difficulty in crossing the technology and business gap.

Image Source: Williams Innovation

Take an example of a mega insurance company's recent comprehensive migration of its home-grown database reported by InfoQ. However, it involves back-end technology migration and transformation. Its purpose is precisely to bridge the gap between technology and business, reflecting the insurance company's need for change at the front-end business level and its desire to drive business development by improving underlying data capabilities. This reflects the insurer's need for change at the front end and its desire to drive business development by improving its underlying data capabilities.

InfoQ reporter recently contacted the AliCloud team responsible for implementing this project to gain a deeper understanding of the story behind the insurance company's unprecedented migration project and to provide more inspiration on exploring the value of data in the financial industry.

All driven by business scenarios

Image Source: Insperity

Before the migration of this home-grown database, the insurance company had nearly 100 online Oracle databases in use for its internal operations, so it was certainly a huge undertaking in terms of migration workload alone.

At the same time, the unique nature of the industry also imposes stringent requirements on business continuity and data accuracy for financial institutions. Although the frequency of transactions and short-term concurrency of insurance is not as great as that of banks and securities, the complexity of the business and the length of the call chain are no less than that of banks, and how to ensure stability during a full migration is another major challenge.

So what motivated this insurer's determination to make changes even in the face of these obvious challenges? Liu Weiguang, Vice President of Alibaba Group and General Manager of AliCloud's New Finance & Internet Industry Business Unit, told InfoQ that it resulted from a business scenario.

In the past, insurance companies mainly promoted and sold their products through intermediaries such as insurance agents and brokers. The traffic pressure was not too great even during big festivals such as the "red-opening" season.

However, with the development of Internet insurance, insurance companies have started to move from behind the scenes to the front stage, and need to face their customers for direct sales directly, underwriting and claims processing, so the high concurrent traffic will inevitably have a direct impact on their back office structure.

In addition, the changes in the insurance industry have also placed new demands on the insurance technology architecture.

In recent years, the domestic insurance industry has been able to connect and serve its customers by combining insurance products with consumer scenarios and industrial ecology. For example, shipping insurance in online shopping, accident insurance in travel, school-level insurance during the school season, etc., are all products of scenario-based insurance products. This means that insurers may face unpredictable and unexpected business volumes from scenarios and ecologies and must be prepared for this.

On the other hand, in this process, the insurance business model has become more fragmented from simply buying and selling directly in the past. Consumers' purchase of insurance products has changed from passive to active, requiring insurers to provide more refined insurance products that meet the needs of different scenarios and can attract more customers, not just the traditional ones. This also requires insurers to offer more sophisticated insurance products that meet the needs of different scenarios and appeal to a wider range of customers, not just traditional generic products. Behind this is the need for a sufficient volume and dimension of data to support and inform it.

"But the problem with the traditional technology architecture is that, firstly, it is concentrated, and if you want to expand it, you have to go vertically to add hardware, which limits its ability to scale and cannot cope with sudden traffic. Secondly, it is closed and difficult to support open business scenarios. Thirdly, it is closed, making it difficult to support open business scenarios. Furthermore, the management of this insurance company's original hundreds of databases is fragmented, making it difficult to control data resources and achieve real-time feedback."

—— Zhang Chi, Vice President of AliCloud, Vice President and Chief Architect of the New Finance Division, pointed out.

The different between traditional architecture and virtual architecture

Open and distributed architecture can be "raised and lowered at any time" according to business needs, so it can solve these problems that traditional architecture cannot solve.

The "pitfalls" of technology migration

Image Source: Mo Khan

However, as mentioned above, the comprehensive localization of the financial database is unprecedented in China. Both for the insurance company and the AliCloud New Finance Division team, the entire technology migration process is self-explanatory, so it is inevitable to step on the "pitfalls".

As we understand, the insurance company advocates a migration path from the periphery to the core. The first scenario to be tested was the insurance company's relatively low frequency and back-end claims business.

The first problem surfaced in the process, the insurance company's underlying technical architecture was still IOE, and the entire environment was not fully prepared for the launch of a distributed database, so two weeks after the first core went live, there were several minutes of blocking in the pooled query system.

Although this was not a "fatal" problem in a large migration of nearly 100 business systems, the negative impact of such a small fluctuation early in the project was a confidence shaker. So, for the project team, the only way to respond was to be quick and to respond in the fastest possible time. Zhang Chi said that in the end, after troubleshooting, they could fix the relevant faults within half an hour.

The second problem occurred during the migration of the legacy core system. The insurer's legacy business core system used a lot of Pro*C (SQL embedded C programs) and Tuxedo (Oracle middleware for distributed transactions) for policy process processing, which meant that the new database architecture had to bypass these two technologies or make a proactive adaptation.

Liu Weiguang told InfoQ that the project team ultimately chose to do compatible with the technologies. "Because it's not just this insurance company that will have to do database migration in the future, there are probably many other domestic companies and organizations that may encounter the same problem, and as trailblazers, for us, sooner or later, we will have to go down this road."

The third problem arises regarding data accuracy, which is also a "hard problem" for financial institutions. As "real money" is behind every business, involving the most critical information about the policy and the customer, any inconsistency or inaccuracy in the data is a huge business "accident".

According to Zhang, the project had been underway for nearly seven months when the project team discovered that some data did not match up while migrating the old cores in the three northeastern provinces.

This left the project team with a new "multiple choice" question. That was whether to do a rollback. If we didn't do a rollback, the data errors could affect the subsequent business; if we did a rollback, it would mean that all the previous work would go up in smoke, which would be a huge blow to everyone. To ensure the stability of the insurance business, the project team decided to revert the whole project. It took about a week to identify the cause of the problem before continuing the work.

Much of the work was overturned and restarted during this process, but experience and confidence were also built up. As a direct result, when the same problem was encountered again, the project team could be more comfortable and quickly identify a solution.

Of course, these 'pitfalls' are only a cross-section of the project, and the migration process has been full of interruptions due to the difficulty of the project, some of which are technical and some of which are not. But the practice also brings real knowledge. By allowing complex systemic projects to be coordinated and regulated by the organization and allowing the number one person in each line to take the lead and make the decisions, the project team members are deeply influenced, and the project is accelerated.

Breaking down "departmental walls" to deliver the value of data

Image Source: Lead Change

In December 2021, the People's Bank of China issued the Financial Technology Development Plan (2022-2025), proposing key tasks such as improving the FinTech governance system, fully releasing the potential of data elements and building new digital infrastructure.

To realize the transformation between building data platform infrastructure and unleashing the value of data elements, there are still some key actions to be taken.

First and foremost is the shift from divide and rule to unified data management. "When we were serving some insurance clients, we found that they had completely different forms for a particular client, a product ID. The forms came from different systems, and the forms, fields and definitions could be different. But when the business revolves around scenarios, insurers will find that information sharing is a basic prerequisite." Liu Weiguang points out.

In this case, companies must first break down departmental walls and transform their data systems accordingly to allow data to be precipitated into assets and unified to do management rather than being a separate report.

The key to this theory is to realize the "workplaceization" and servicification of data. The key to this theory is to realize the "workplaceization" and servicification of data. In other words, any business person can directly access the model algorithms and data capabilities of the middle and back office (such as the portrait of a certain customer, etc.) through one service when they have any needs to help them do analysis and make decisions.

It is worth noting that there are still two core issues to be solved. On the one hand, it is the understanding and abstraction of the business because the data needed and the models that may be used are different in different industries and scenarios. On the other hand, the clarity of data attribution, including details such as who owns the data, who builds it, who produces it and who is responsible for it, etc., should be clearly defined. Long Xiaoping believes that to solve the first problem, business departments must participate, collide and integrate with technical staff. Top-level design and clear authority and responsibility are needed to solve the second problem.

This is why digital transformation must be a hands-on project because it is not just an IT upgrade but a change in all aspects, requiring the cooperation of all staff.

So, how do you get "everyone" to participate? Liu Weiguang stressed that it is important for all people to feel the power of digitalization and benefit from it.

For example, we see that today financial institutions are giving the best new technology and experience to their customers, but the internal staff office experience is very backward. From internal office approvals and document flow to accessing IM for internal communication and logging into the business side to check customer messages, all are scattered in different systems, and even initiating internal video conferencing, email systems and group messaging requires switching back and forth to get through.

Back to the question we mentioned at the beginning. Why do many companies lack data utilization? One of the reasons could be that the data is not easily accessible or the chain driving the business is too long, resulting in many employees not being directly empowered by the data and the value of the data not being fully exploited.

Behind this is a problem with the strategic positioning of digitalization. Many companies treat digitalization as a technology project transformation rather than an overall capability system, focusing only on the external customer experience but lacking attention to the internal employee experience.

In other words, to deliver the value of data, in addition to basic platform capacity building and smooth departmental collaboration and interaction, it is also necessary for enterprises to build a unified perception of the concept and value of digitalization from a top-down, internal to external level.

Digitalization is a long-term project that we have to consider the cost of time

Image Source: Desuvit

Strategic positioning is, therefore, very important. It is like a sniper rifle's scope, which calibrates the target and prevents misfires. In this regard, Liu Weiguang summarized three points of attention for enterprises in formulating digital transformation strategies.

Firstly, you cannot blindly learn and imitate but must plan the goals and path according to your characteristics.

"Some enterprises will first make changes at the level of organizational structure, some will start with the transformation of the underlying infrastructure, and some will also need to improve their brand image first. The first question to figure out is what exactly are the weaknesses and pain points of the business, and what are the strengths and opportunities? For example, some large financial institutions do not lack a customer base, so the immediate goal is to serve their existing customers well and do refined operations. For small and medium-sized financial institutions need to seek incremental growth first and make the customer scale up."

Liu Weiguang told the story of a city agricultural and commercial bank: this bank mainly serves local farmers and fishermen, and many demolished households have a very high potential for savings. The bank's and its depositors' obstacles were the corporate image and brand appeal. As a result, Liu and his team suggested that the bank should start by improving the functionality of the mobile app, using it as a local dialect service and as a cultural bearer for the ethnic group, which gradually increased the bank's influence in the local area and increased the number of customers.

A noteworthy detail is that the bank's staff can switch between internal and external interfaces with just one click using the mobile app. For example, suppose you open the APP anywhere and switch to the internal staff interface. In that case, you will be able to help customers with their cards and business, and you will be able to directly feel all the processes and experiences that need to be improved during your daily use of the APP as an external user. Liu Weiguang pointed out that this is a manifestation of digital capability services empowering every employee.

Secondly, companies must ensure a strong mapping relationship between business and technology strategies.

This is an important prerequisite for facilitating effective communication and collaboration between business and technology departments. In the past, the division of labour between business and technology was very clear, and technology departments had a limited voice. However, digital transformation means that companies must invest in technology and, at the same time, make changes in their corporate DNA, based on which they can then consider the relevant talent to be brought in. In other words, the company's DNA is the basis for attracting talent. In contrast, talent is the key to implementing the company's digital strategy, and one cannot be without the other.

Thirdly, digitization should be considered both a long-term project and a time cost.

A relatively unanimous consensus on digital transformation in the industry is that it cannot be achieved overnight. However, this does not mean that enterprises do not need to consider the input-output ratio but instead should focus more on dismantling objectives and quantitative assessment.

"Digital transformation in enterprises should be about setting a small goal at one stage, rather than a big direct multi-year goal or multi-year plan. In this process, it is necessary to assess the personnel input and technology input, and then when the time comes to proofread whether the target has been achieved, what the specific progress is, what the reasons for not achieving it are, etc. We keep approaching digitalisation by constantly iterating and updating the strategy and technology." Liu Weiguang explained.

All in all, from small data capacity building and data value mining to digital transformation, they are not mere technology projects or a combination of several technology projects. Still, they require companies to do good layout, deduction and practice from all aspects, such as culture, process and technology, and gradually achieve breakthroughs in the process step by step.

Interviewers:

• Liu Weiguang: Vice President of Alibaba Group, General Manager of AliCloud New Finance & Internet Industry Division.
• Zhang Chi: Vice President of AliCloud and Chief Architect of the New Finance Business Unit.
• Long Xiaoping: Senior Specialist, AliCloud New Finance Strategic Client Department.
• Ren Zhenzhong: Solution Architect, AliCloud New Finance.