Retail data view (2): how to understand membership ID


Preface:The traditional retail business has entered a bottleneck period, and the concept of “new retail” emerges in endlessly, dazzling: Omni channel / unmanned store / Unmanned Container / New Ecology (HEMA fresh / millet home / tmall store), etc.; all kinds of technical concepts are even more confused: oneid, data center, user center, unified member platform and idmapping, etc.The main content of this paper is to clarify what is “membership ID through” and application scenarios, so as to help enterprises better complete the integration of omni-channel members, multi brand members and the accuracy of user operation.

Note: the “members” mentioned in this article refer to Pan members, including users, customers, potential customers or customers.

I. what is membership ID through

There are two business scenarios for opening member ID:

  1. Get through the same identification ID of each business system to form a member ID, for example, get through the members in brand a CRM and brand B CRM (through mobile phone number) to form a unified interface service, and then provide a general service interface for the business system to call, such as: user registration, user information modification, point consumption, coupon sending, etc.
  2. Get through different identification IDs of various business systems to form member IDs. For example, through the user’s consumption behavior on Taobao, click browsing behavior on wechat and offline store consumption, form a full-scale user “hologram” to provide data support for user operation. Common business scenarios include commodity recommendation, personalized activity push, etc., which are mainly used to improve the repurchase rate or single connection rate of old customers Mark.

II. Business scenario 1: get through all business systems with the same ID to form a member ID

Abstract work of traditional business sharing module requires detailed investigation of user related operations and abstraction into unified data structure, such as the following scenarios and corresponding detailed business subprocesses:

Retail data view (2): how to understand membership ID

The next step needs to be abstracted into a unified data structure, which may be as follows:

Basic information of members:User ID, wechat, mobile number, consumption password, name, gender, receiving address, age, points, coupons, labels *, etc

Registration record:Registration time, channel and detailed address

Binding record table:Binding time, password mode, mobile number, openid, etc

Member change record:Member account number, account change time, participation content, account change information, etc

Member coupon information:Member ID, mobile number, coupon number, coupon amount, coupon name, number of coupons, validity period, use requirements and other relevant information

Technically, the above information will be put into a relational database, and a highly concurrent service interface will be provided for the upper business to call: such as member registration, member information query / modification, coupon issuance, etc., which we often define as a user center or a unified user platform, in the same way to meet more business needs of enterprises, there will be order center, commodity center, etc., traditional enterprises in In architecture design, this layer will also be defined as data middle platform.

III. business scenario II: get through different business identification IDs to form member IDS

First of all, let’s know what identification IDs are available. Common basic IDS include mobile phone number, ID card ID and email address; unique ID of user in information software system, such as user ID; cookie ID most commonly used in PC era; various device IDs introduced by the emergence of app, such as IMEI / IDFA and MAC, and open ID generated by wechat; I generated by rapid development of face recognition in recent years D, such as face? ID.

Now let’s classify members from another dimension: identifiable members, reachable members and descriptive members, as shown in the following figure:

Retail data view (2): how to understand membership ID

  1. “Identifiable member”: judge whether a natural person can be identified. All the IDS listed above are.
  2. “Touchdown members”: to determine whether members can reach members through ID touch. The related ID will include Email (mailbox), mobile (short message), imei/idfa (advertising), MAC (advertising), user_id (APP station letter), open_id (public content push).
  3. “Describable member”: judge whether ID has label information that can be associated to describe user’s portrait, and describable member is the data base for enterprises to carry out refined operation of members. Enterprises should use all information-based means and label design (how to design labels, please refer to “Lan college retail data view (1): how to spend 30 minutes to become a label design” Da Ren “, Expand the scope of descriptive members, but there is also a very important step: how to get through the tag information associated with different identification IDs and form a member ID, which is often called oneid, and idmapping is the process of describing the transformation between oneid and other IDs or forming oneid.

The simplified structure of oneid might be as follows:

Retail data view (2): how to understand membership ID

The generation of the above scenarios may be that oneid1 is online Taobao consumption, oneid2 is offline wechat payment consumption, the two are connected through the same mobile, oneid1 and oneid2 may be the same person or a family, and through idmapping, the tags behind oneid1 and oneid2 can be connected, so as to make the user’s operation more real and three-dimensional.

Note: the real business scenario will be more complex. The conversion of ID and ID needs to consider the probability. For example, mobile1 and mobile2 are mobile numbers obtained from the harvest address. The probability of mobile1 occurrence is 70% (10 orders), and the probability of mobile2 occurrence is 30%. When oneid1 is converted to mobile number, the probability of mobile1 is the highest, and mobile2 is converted to oneid, and the probability of oneid2 is the highest, Oneid1 is only 30%.

Summary: Based on the above introduction, I believe you have a basic understanding of the basic classification and application scenarios of “membership ID unblocking”. In the actual project implementation process, there is no conflict between scenario 1 and scenario 2. Scenario 1 belongs to the traditional information construction category, and scenario 2 belongs to the big data application construction category.

Brief introduction of the author: tiejiaoshou, 10 years of experience in data related, has been engaged in DBA, data warehouse and solution in Telecom and Alibaba, mainly experienced the core coder of Alibaba to IOE and oneid, and is currently engaged in the solution of retail industry.

[for more information, please visit datalan community, the first data exchange community in China (…)]