no.6 edit

Do you know the DMP data management platform?

DMP is Data Management Platform, a data management platform. In recent years, with the popularity and popularization of big data and program advertising, the DMP market has grown rapidly and is expected to exceed 200 billion yuan by 2020 (Source: Mob Research Institute).
DMP's core idea is to record each individual user's "interaction traces" on different marketing contacts, based on which to distinguish the characteristics of different users, and to provide targeted marketing strategies for different characteristics of user groups, or to export these groups as segmented audiences to other marketing agencies.
Part 1: What is DMP? What are the features?
DMP is a comprehensive data collection, processing, integration platform, absorbing a variety of data source data, user-based units, cleaning, organizing the formation of structured data sheets, and user label calculation, in order to be able to accurately describe a variety of users. DMP does not have any marketing nature and functionality, it can be combined with DSP (Demand-Side Platform, demand-side platform), CMS (Content Management System, Content Management System), CRM (Customer Relationship Management, Customer Relationship Management System) and other products, there is some data flow, the core of DMP is to process the data. Powerful DMP is the data brain of the Internet age and the foundation of all marketing.
The function of DMP
Precision marketing, ad serving, personalized recommendations, and other data analytics applications
DMP core features
1. There are many data sources, including in-enterprise data (first party), cooperative data (second party), external third-party data;
2. To the user (or equipment) as the basic unit, the integration of all parties data, data cleaning, structured storage;
3. The structured data is analyzed and the algorithm is run, and the user is multi-dimensionally qualitative and labeled.
4. Users can extract the desired user base through DMP, and downstream marketing products can access and extract data directly through interfaces.
Part 2 : DMP's infrastructure and data processing processes
A function module required by DMP is: (1) data acquisition inlet, (2) data cleaning and integration module, (3) label calculation and mining module, (4) labeling and label display module, (5) user group module. Among them, (1) (2) (3) is the basis of DMP, determines the quality of a DMP.
Part 3 : DMP classification
1. Core Internet factory owns DMP
Such as Ali's mother's Damo disk, Tencent's Wide Point Pass, Baidu's DMP data services; The core Internet factory has absolute user traffic advantage, so the amount of data accumulation is very large, but the data between the major factories is relatively closed, not circulation and sharing.
In addition, the major factories of DMP will also be based on their own business to do some deep customization, such as Ali's Damo disk is for e-commerce scene DMP, versatility may not be high.
2. Advertising technology companies own DMP
This kind of DMP is mainly built by advertising technology companies based on the expansion needs of DSP and other business, directly for business services. The U.S. market is very segmented, with many dedicated DMP companies, the Chinese market is highly integrated, DMP demand is often associated with DSP/SSP/Trading Desk, and advertising technology companies have their own DMP to meet this consolidation need more directly.
3. Advertisers own DMP
This kind of DMP data is relatively closed, data security is better, but the corresponding circulation is poor, data mining and under-use, usually need to access other data platforms, data fusion and circulation, in order to better play the value.
4. Independent third-party DMP
With the development of big data technology, precision marketing has become a necessary option for enterprise user growth. As a result, a number of third-party data providers, known as third-party DMPs, have emerged that serve enterprise customers. Third-party DMP is relatively independent and professional, and enterprises can quickly obtain the desired user data. But at present, the domestic market for data transactions, data exchange has not yet the corresponding implementation standards, data security can not be well protected.
Part 4 DMP's main application direction in marketing - user portrait
There are two types of definitions for user portraits: User Persona and User Profile. l User Persona: is a typical user that product design and operators abstract from the user base. For example, during the user research phase, the product manager learns about the commonality and differences of the user through questionnaires and customer interviews, and summarizes them into different virtual users. l User Profile: Outputs a collection of labels that describe users based on each person's user behavior data in the product. For example, guess the user's gender, where they live and work, their hobbies, which star they like, what they want to buy, etc. :: The user portraits mentioned below are All User Profiles.
The relationship between the user portrait and the DMP
DMP, as a data management platform, outputs a series of labels, or other desired results, after data is entered. The user portrait is entered into the user data to the DMP, which then outputs the user label. User portrait is the main application direction of DMP in marketing. Common applications for user portraits include personalized recommendations (e-commerce, information products), wind control, forecasting, etc. In fact, DMP can not only output user labels, but also output other labels, such as input articles, output article labels and so on.
How the user portrait does it
1. Basic data collection: collecting static and dynamic data from users during the use of the product/service;
2. Behavior modeling: based on the user's basic data, behavior modeling by technical means;
3. Build portraits: Through behavior modeling, a series of user tags are output, and each user's label can form a collection that represents the user's characteristics.
A sound labeling system can help advertisers do multi-dimensional, all-round user portraits and insights, so as to achieve accurate targeting of advertising. Advertisers can use the label system to freely choose, combine the characteristics and behavior of the audience, accurately target the target audience for advertising. At the same time, advertisers can also use the label system to their own seed population data for in-depth mining, according to the characteristics of seed population to expand more similar groups, expand the potential user base for advertising.
Is your supply chain a smart supply chain?
5 minutes to learn about machine learning
Big Data: Basic Concept