How do I distinguish service boundaries and how do I segment them from system data ports when the service boundaries are known...... These problems have become a challenge for many product students and even data students. In this article, the author shares the idea of cutting and docking the business system, let's take a look.
In the current Internet environment, there is no longer the previous state of frantic expansion of various business lines, and the overall business has begun to gradually shrink to the familiar business of each company, and it is relatively rare to develop new business lines.
As you focus on your core business line, you will gradually find that in the past crazy expansion stage, each business line was not clear enough, resulting in a fuzzy business data and boundary between the upstream and downstream departments of the unified business line when focusing on the core business, and the handover is not clear, or there are certain business logic conflicts in some parallel departments of overlapping business.
When the business environment is good, this is not a problem, when the business ports are tightened and cut, every inch of the remaining core business line is a battleground, in such an environment, it will inevitably cause a certain amount of trouble to the data and product builders, how to distinguish the business boundary, how to do the slicing from the system data port in the case of the known business boundary, how to find the appropriate sharding point and sharding logic in the case of unknown business boundary lines, which is the challenge point of the colleagues who do product or data in the middle.
In the segmentation of boundaries and the circulation of data, it is very difficult for product and data personnel to carry out core business transformation under the condition of keeping the current business structure and key data node data unchanged, which requires a deep understanding of the current business, determine the bottom table and structure of each port that are not affected, keep it unchanged, add some judgment rules and policies, and improve the transformation of data flow.
In terms of the environmental situation I have faced recently, I would like to share with you the idea of cutting and docking the business system.
1. Demand analysis
Let's take the overdue behavior after the user enters the risk control line to share the life cycle management from the user's loan to the arrears.
In general, users go through the following life processes:
Taking the current lending behavior of Internet finance as an example, because the era of early investment in drainage resources to pull borrowers has passed, the current is more about the development of existing users, mainly focusing on the borrowing method of revolving credit similar to credit cards, that is, maintaining a stable credit line within a certain period, users can revolve and repay within the credit cycle, as long as the principal does not exceed the credit line, they can always be recycled in repayment and borrowing.
Due to the existence of revolving loans, the relationship between users and institutions has changed from single borrowing to revolving multiple loans, which is more reflected in the sticky value of Internet financial institutions.
In this case, it is necessary to manage the user life cycle of the core business of different business parts, clarify the key behaviors of the beginning and end of the user's life cycle under different business cycles, and determine the time point of the user's life cycle through the key behaviors.
Institutions will also pay attention to users who are in arrears but can repay after collection or legal collection, so the life cycle management of users is no longer a single flow.
For legal reminders with higher costs and more standardized means to dispose of overdue, after dealing with relatively high aging customers, the user's aging drops to M3 and below, whether it is necessary to roll back to the collection follow-up
For users who have already incurred overdue payments and completed the overdue part of the payment, is it worth re-entering the stage of credit access to score the credit amount again, and allowing the user to initiate another loan?
All of the above involve the dismantling of data flow and business scope of different business systems
At the same time, because the business itself of legal reminder and electric reminder is not completely front and back, for mutual finance companies, this is the difference between different disposal methods of overdue assets, legal reminder is more forceful, more later, electric reminder is more early, more cost-effective, and ROI is higher, so the similar or overlapping parts will have certain changes due to the disposal expectations in different periods, and more cost-effective collection will be adopted in the cycle of more relaxed supervision, and litigation and other legal reminder methods will be used to recover overdue arrears in the period of stricter supervision and better judicial disposal environment。
Therefore, it is inevitable that the same user will scroll back and forth in different business departments or business lines, and even within the same department, it is inevitable that there will be rolling adjustments of business data before different services.
Therefore, it is necessary to define the boundaries of the business scope to a certain extent, and complete the transformation of the system on this basis.
Second, the demand design
Before designing, it is necessary to clarify at what point the essential changes in the mode will be handled, and use the essence as a slicing point to give both ends of the business a certain degree of autonomy.
To put it simply, you can try to establish an intermediate state of a user in the middle of two services, and there are several states of intermediate users:
- Targeted push collection
- 催收定向推送法開催
- The parties are temporarily unable to dispose of the status
- If the overdue part is fully settled, the front-end risk control credit status can be pushed
The screening of the scope and the establishment of intermediate states can help both the business and the front-end risk control of the two parties to make certain cuts to their respective business lines and the users who are about to enter their own business lines, and at the same time have a certain understanding of the user situation and quality they are about to face.
Under the premise of establishing an intermediate state, it is necessary to manage the life cycle of users within their respective business scopes.
For collection, most of the collection is in the form of orders and account periods to follow up, and rarely use the way of user return, while the follow-up form according to orders and account periods is more convenient and fast for collection, and the immediacy is stronger.
Although different orders may be different ways to resolve loan disputes, they will be combined to avoid splitting multiple orders to sue in multiple places under the case of high disposal costs, and merging orders to users at the same time has a better management effect on the overdue behavior of users for a longer period (users who enter the legal reminder usually have a longer period of complete settlement, and it is more intuitive and effective to follow up the user's repayment behavior after having the concept of returning) 。
At the same time, for the company, the users who try to drain as much as possible in different ways are usually processed at the front end according to the main key of one user and one ID card, so that the credit line can be accurately located on the person, and avoid filling in different backgrounds on the user's different loan orders to maliciously apply for loans.
Therefore, it is best to use the form of the user to establish the intermediate state, and for the collection, the construction of the lost card and the user's post-loan behavior analysis card can also be observed and followed up in the user's dimension, which has certain value for the three parties in the business line.
There are two key states in the follow-up of collection, one is the beginning state of overdue, and the other is the final state in which the current user can not continue to deal with the collection, and the definition of this final state has a certain relationship with the overall management and cost considerations of each institution.
However, the basic logic must be the overdue aging, when the overdue aging of a user's order reaches a certain stage (such as M6 and above), it means that the order is overdue for at least 150 days or more, and the user's linkable status declines.
In this case, for legal collection, the beginning stage of the user must be the beginning of the state of collection can not be followed up normally, and the definition of this beginning is closely related to the evaluation of the user of the collection, which requires the collection to be in the end of the case push state, the user is labeled with the corresponding label, for example, the loss of contact into the legal reminder, the overdue account aging is too high to enter the legal reminder, the early litigation into the legal reminder, the end of the previous business line business, is the beginning of the next business line business.
In this case, the data inherited from the front-end is particularly important, with key nodes to do data segmentation, retain the key state of the user, display in the intermediate user pool, after entering the legal reminder, the legal reminder can complete the matching of assets according to the initial state of the user in the collection, and determine the optimal disposal plan of the asset, whether it is litigation, asset sale, mediation or other feasible disposal plan.
After experiencing the disposal of collection, the simple recovery method can no longer quickly recall the overdue part, which requires further screening of the type and quality of the asset, in order to complete the recovery of non-performing assets at the fastest speed, at this stage, it is also necessary to consider the depreciation of the non-performing asset package over time, and in the case of discounting, you can try to make a certain reduction or exemption on the overdue assets to avoid the shrinkage of the asset value caused by the longer the asset lasts.
From the perspective of recovering assets as soon as possible, users who are overdue in legal reminders can be released to the intermediate user pool, and the risk control combines the disposal cycle and duration of collection and legal reminders, and the front-end risk control input port further evaluates whether the user is worth the credit access process.
After experiencing the disposal process of collection and legal reminder, the user's life process is gradually clear, which has a certain value for the company's middle office department to evaluate the user's life process, and the process itself rolls out from collection to legal reminder and then from legal reminder to the design of front-end lending products, which can clearly understand the possible repayment time and bad debt trend of users who may be overdue, and the whole process evaluation of user behavior has a certain gain effect on the iteration of the front-end risk control model. At the same time, it can also screen some users with strong stickiness who will still borrow on the platform after going through collection and legal reminders, and achieve more accurate credit stratification for users in the credit port according to the characteristics of strong sticky users.
3. Data realization assessment
In the construction of the above business requirements, the difficulty lies in the construction of the data flow, and in the collection port, the status of the first overdue (the first overdue of the user/the first overdue order) is often paid attention to, so as to avoid the user from losing contact quickly after the first overdue.
Collection often in the first more than a large number of manpower, so the first stage of user behavior records and collection behavior records is very important, in the user gradually overdue flow to the legal reminder stage, overdue behavior increases, the behavior of the collection decreases, to the legal reminder, the legal reminder behavior disposal is important, but the user generated by the joint behavior and repayment behavior record is more important, this bilateral behavior record needs to consider the cycle. In the early stage, a large amount of data may be recorded, but in the legal urging stage, what should be recorded is the time when the behavior is generated, and the flow of data needs to be cut at the time node of the specific behavior.
For business performance statistics, the most critical may be the time of collection and the time of collection, and similarly, for legal reminders, it is also necessary to divide the user's input time and outgoing reminder time, but at the same time, like the business logic that has been determined before, there may be users who have gone through post-loan, legal reminder, At the same time, it is also possible to enter the overdue part of the legal reminder into the intermediate user pool again, and at this time, it is also necessary to choose whether to enter the legal reminder stage or the collection stage according to the rules.
In this case, it is very important to slice the time of the user's life cycle, which can avoid the simultaneous existence of one user in collection and legal reminder at the same time, and can also cut the performance observation of the two parties, avoid the situation that the business distinction is not clear, and at the same time, it can also avoid the situation that one customer experiences two disposal methods at the same time and generates customer complaints.
At the same time, the establishment of user life cycle time slices can help the collection and legal reminder to determine the user's last state and historical status at the time of the user's entry, and whether the user has been followed up in the collection and legal reminder in the past, which has a certain efficiency optimization and improvement effect on the disposal of non-performing assets.
epilogue
For the internal business products of Internet companies, how to design and update to adapt to the current market changes and business changes is very important, in the current threat of the external economic environment, how to improve the efficient cooperation between various departments, to avoid the waste of redundant human resources and product resources is a topic that every company needs to think deeply.
Columnist
An Chen said that everyone is a product manager columnist. Focus on financial risk control and explore the direction of risk control in different economic environments.
This article was originally published on Everyone is a Product Manager. Reproduction without permission is prohibited
The title image is from Unsplash and is licensed under CC0.
The views in this article only represent the author's own, everyone is a product manager, and the platform only provides information storage space services.