How to create a data governance closed loop?Insurance industry as an example

Editor’s Guide: Data Governance refers to a set of management behaviors involved in the organization. The insurance industry is the earliest, and the most complete perception of the value of the data. Therefore, their demand for data governance is very strong. The author of this article will create a data governance closed-loop deployment analysis, I hope to help you.

I have a lot of fun. Just got a consultation work for the major data project of the transportation industry, and there is an advisory business for insurance industry data. I am still very embarrassed at the beginning, afraid that crossing the industry, the experience is not good. It was found that the insurance industry data governance is still the set.

At the same time, I still pay attention to data, this project is too comfortable! Today, I will share the experience of the insurance industry data governance project. I think the data governance in this industry can be used as a benchmark.

I. Insurance customer data governance background

The insurance industry is the earliest and most complete perceived the perception of data value throughout the financial industry. Therefore, their demand for data governance is very strong. This is the industry attribute, and of course there is also a policy requirement for the China Insurance Regulatory Commission.

Insurance customer data is a very important data resource in the insurance industry. Many insurance insurance budgets are based on customers, plus other messy data together.

Therefore, the management of insurance customer data can basically be divided into three categories:

Business requires data management requires policy needs

Business needs is easy to understand, the basic information of the customer collected from the hard work is filled out, the contact can not contact people, this insurance bought? The more full data, the more accurate, we can “grade the group to set the prior guest”, in the Internet, it is to do the customer hierarchical, precise marketing, final insurance, take the year-end!

Data management needs can also be understood, after all, various subsequent data analysis, the calculation is based on customer data. Customer data is wrong, and that analysis is definitely deviation.

Don’t explain more about the policy needs? Who can I listen?

The purpose of insurance customer data governance is also very simple, just enhances the authenticity of customer critical information. Generally, the three elements: ID number, customer name, customer mobile number.

Second, the analysis of insurance industry data problem

Although the insurance industry attaches great importance to data, many years of rapid development of insurance business has indeed brought a lot of data issues.

All local branches are basically the politics, who make customers a life root. The salesman is even more, even some people deliberately hide real customer data, that is, afraid that others will take their customers.

This has a very funny scene: performance is completed, the premium is rising, but the customer phone is not able to do it, and it is not a real person. I have seen an insurance agent to read information, answer the phone across the phone, and I don’t know if it is helping.

So, go to the organization level, down to the business execution level, there are different levels of data accuracy of different levels. Do you say this data is good?

It is said that the situation in this insurance industry is just the same as the second-hand housing! The government also called for “true listings” for many years, and there is no solve the market, and finally the shell is governed. I think an insurance company has to learn from the ACN rules of the shell, and maybe it can be able to subvert the insurance industry and become an insurance giant.

Third, the insurance customer data governance process

The process of data governance in all industries is also as long as this:

Of course, the above is the actual version. When executed, of course, according to the logic of project management, first build the overall plan, set the job, and promulgate the management method. Midway has to supervise, control, but also set several milestones to ensure that the tasks are completely prepared on time. However, those are all about project management, here is not described here.

The first two steps of data governance are definitely data collection, data processing and analysis. These two steps are basically all data engineers working.

The data collection is actually all of the client data in various systems such as platforms, Call Center, CRM, various business systems (underwriters, claims, farms, etc.), complaint systems, etc.

Data processing is actually to be a customer’s return and use technology to manage data. Put the technology can be cleaned, match the data on the matches, such as remove the +86 of the mobile phone number, the space in the middle of the customer name. Then set various rules to perform validity determination, such as: three factors data lack, irregular, and test.

True to data governance, in general, it will be divided into two parts:

Stock customer data cleaning; incremental customer data quality control.

1. Stock Customer Data Cleaning

To deal with the stock customer data, organization, you need a comprehensive data quality verification, and distribute the results to various local branches. All local branches get a list of issues, basically, to analyze the causes of problems, and formulate cleanup schemes according to different factors.

If it is technology solved, then it is better. If the technology is solved, the salesman is arranged to make the stock data cleanup work. In summary, all branches, according to questions data list, control data quality rectification standards and tasks, each schedule data cleanup work. Because the technical rules have been clear, this part is basically cleaning. This workload can be imagined, the most powerful, most laborious. 2. Incremental customer data quality control

For incremental customer data, it is better to do. Let the following strict implementation. Insurance is a strong control scenario collected by a customer data. After all, it is closely related to individual money.

The method of strictly enforcement is actually controlled. Anyway insurance company will contact customers in sales, underwriting, claims, consulting, and services, and customers have repeatedly inquiry and checking information every session.

After the salesman re-confirms the customer data, the customer resource system will begin to move, according to the latest customer data, repeated comparison, confirmation, update the customer data, do data, and data return, data. Is this customer data effectiveness not going to a little bit?

3. Data Governance Support System

The salesman is being manually cleans the data, and we are not idle here, and all kinds of monitoring are got! The monitoring large screen should be placed on the leadership desktop, and the mobile phone is listed; various monitoring daily reports, weekly reports, monthly reports have been designed, regularly distributed; various branch data validity list has been arranged. This is the efficiency, I’m going!

I have a problem out, some people say that data analysis is not used, and the actual value is not seen. It doesn’t mean that the data is changed to money. Sometimes, a leaderboard can be multiplied by the improvement efficiency. Is this not worth?

In addition to data drivers, it has to be driven by organizational drive. Therefore, the supervision team will also be formed, on the one hand, progress monitoring, on the other hand, supervise and guidance on the problem of problems, and assist them for data governance.

Of course, KPI drive is also very important. The final results index also have to be released to form the closed loop of the final data. In general, the real rate of the customer is true, and the real customer renewal rate is the final results indicator.

Finally, I would like to give people a prize, and I have worked hard for half a year. This is still very important, or no one is active next time!

Fourth, summary

Data management is a topic of an older talk. The most difficult thing in data governance is how to promote organizations and implement this matter. But this thing has natural driving force (policy, value) in the insurance industry, and there is natural resistance (personal interests and organizational interests).

The insurance industry data governance workflow is roughly related to the data governance process of other industries, and it is to collect data first, and then use technology to clean, return to. This part can actually use the previously mentioned one ID, the efficiency will be high.

After the technical process is finished, the remaining invalid customer data will be handed over to manual treatment. Headquarters does data quality analysis, process monitoring, supervision, and all branches completely implement it. Using various methods and means such as KPI, list, drive all people to achieve data governance.

After the data is updated, the system will then make a customer discrimination and data return, so the data quality will continue to improve.

Individual understanding, the most difficult part of data governance is not technology, but how to transfer the strength of the entire organization. In fact, technology can do very little, most of which requires common efforts of various departments such as business, personnel. I have been more relaxed in the management of government industry data, but it is very difficult in Internet companies. Later, I summed up, nothing more than the reason for the organization’s driving force.

Author: big data architect, public number: big data architect, Director of China Pharmaceutical Country, Good at Bi, Digital, Data Mid-Taiwan Product Planning