The whole process docking of consumer finance-credit product API

There are two kinds of credit, pre-credit and normal credit:

Pre-credit: The lending platform relies on its huge user base and accumulated user data, such as behavior data, transaction data (credit platform-loan order post-lending data, e-commerce-transaction order data, payment tools-payment transaction data), and can evaluate user qualifications and grant users a quota in advance under the calculation strategy of the platform's own risk control engine without the need for users to submit additional information. However, in this way, only relying on the nationwide application of Alipay and the cattle loan business of cattle credit housekeeper tools like ants can users collect enough data for risk control, and most ordinary credit products still cannot effectively adopt the pre-credit method.

Conventional credit granting: At present, the common way is for users to submit various information items, and evaluate the credit products according to the information items submitted by users to get the loan amount.

The following is mainly about the conventional credit method.

As a lending platform with full API process access, it generally has its own standardized API process to connect with credit institutions. Generally, after determining the cooperation intention, the interface file will be thrown to the credit institution first, but whether it can be developed in full accordance with the interface file naturally requires constant communication. According to the standard API document, there is an obvious advantage for the party providing the interface document, which is to save time and trouble.

Common credit links basically include the following basic authentication steps:

(1) Real-name authentication: refers to real-name two-factor authentication, that is, name+ID number to prove that the borrower is not a black account (recorded in the Public Security Bureau);

(2) Operator authentication: through user authorization, capture the data of the mobile phone number filled in by the user at the mobile phone operator, which may involve the basic information of the operator, mobile phone package, bill data, call list, short message list, recharge list, points list, online list, short family number, etc.

(3) Personal basic information authentication: including three parts: ① Personal information (education background, marital status, etc. ); (2) Professional information (basic information of industry and company); ③ Contact information (obtained through authorized mobile phone address book+filled in by the borrower). Here, ① and ② can generally be filled in casually, after all, it is really impossible to verify; (3) Generally, the mobile phone address book is grabbed by compulsory authorization, and then the user fills in the emergency contact (generally used for the borrower's collection after malicious overdue). If the contact person is not his own address book friend, and the relationship between comments and choices is inconsistent (there may be manual review), he will generally fail the preliminary screening;

(4) ID card OCR and live identification:? There are two processes (both using third-party services): ① Check whether the borrower currently applying is alive through several actions [blinking, opening mouth and nodding]; (2) Identify keywords (mainly ID number and name) from the uploaded photos of the front and back of the ID card, connect with a third-party identity information data source (Public Security Bureau) to obtain a comparison face photo, and then compare the living photo with the comparison face photo, and pass it after reaching a certain confidence threshold (similarity), mainly to prevent borrowers from using other people's photos purchased from the Internet or the black market).

In addition to the above basic processes, other authentication items are determined according to the different risk control of each credit product and the different user data required, and the following differentiated data authentication items are accordingly available:

(1) Credit card online banking data: the user manually authorizes the import and acquisition;

(2) Credit card information: the user fills in a credit card and carries out four elements of credit card authentication (ID number, name, bank card number and bank reserved mobile phone number);

(3) E-commerce-Taobao Alipay data: the user manually authorizes the import and acquisition;

(4) Others ...

The more data items users need to authorize or fill in, the higher the churn rate. Therefore, some lending institutions will place unnecessary data items in the withdrawal process, such as binding savings cards.

The borrower fills in all the information, submits the credit application and waits for the risk control. Generally speaking, lending platforms with a little risk control capability will undergo the preliminary screening of risk control (user data will help to gradually train the risk control model), which is also conducive to subsequent self-operation.

After the risk control audit is passed, the credit institution will give users a certain loan amount, which is divided into two types:

(1) revolving line mode, which gives users a certain loan line. As long as the remaining amount is greater than the loan amount of a single loan, the user can withdraw the money within this amount to the savings card within the validity period (generally 1 year), and after paying off a loan, the remaining amount will be restored to the cleared loan amount. Example: The appropriation amount was ￥ 65,438+00,000, and a sum of ￥ 65,438+000 was borrowed and another sum of ￥ 65,438+000 was borrowed, and the first sum of ￥ 65,438+000 was paid off, leaving a balance of ￥ 9,000. But in fact, if there are a lot of withdrawals, some of them are overdue. Although there is a surplus in the quota, it is still impossible to withdraw cash.

(2) Single batch mode, which gives users a certain loan amount, and users can only choose to withdraw all the current amount or not. The validity period of the grant is relatively short (usually 7~ 15 days), and some credit products are allowed to reduce the amount appropriately. Example: The appropriation is 65,438+00,000, and only 65,438+00,000 can be borrowed, not just 65,438+0000. If you need to borrow again after the signing withdrawal or the expiration of the validity period of the signing withdrawal, you need to initiate a credit application again (of course, you don't need to fill it in again if the data authentication items that have been filled in have not exceeded the validity period).

I talked about two kinds of cash withdrawal, and then talked about the common problems in the process of API borrowing:

(1) Binding savings card: Some credit products will put this process in the credit granting link, but it is not necessary. If the user's approval is rejected, there is no need to use the bank card information at all, and there is no more link. After credit products are granted to users, it is more reasonable for users to take a process of binding cards when they manually withdraw money to their savings cards. The card binding process needs to consider the following issues:

(2) Loan interest rate

Theoretically, the borrower's repayment plan calculated by the system when borrowing money should be consistent with the repayment plan of subsequent users after the actual loan is successful.

Only if the loan interest rate is lower than the annualized interest rate of 36% is it compliant. In fact, the interest rates of some large and medium-sized credit institutions plus management fees and service fees are still within 36%, but now the whole industry is in a reshuffle period, and the cost of obtaining customers and operating expenses are increasing day by day, and the annualized interest rate remains compliant. For some credit institutions, differentiated interest rates for users with different qualifications can solve some problems, but there is an upper limit, and there is still some pressure, so some cash loans and even super-interest loans have actually far exceeded 30.

Credit institutions are actually more sensitive to users' post-loan data. The loan platform will generally aggregate the loan records of users in various credit institutions to facilitate users' repayment, so order synchronization is necessary. As shown in the figure below:

The order data can be synchronized, but the detailed repayment plan in the order data (user's repayment plan and repayment situation of each period) is not synchronized by some credit institutions, so the loan platform cannot obtain these data. Under normal circumstances, users must go to the repayment H5 of the credit institution to operate.