Since 2020, the tripartite cooperation between the Commercial Digital Marketing Professional Committee of China
Since 2020, the tripartite cooperation between the Commercial Digital Marketing Professional Committee of China Advertising Association, the Organizing Committee of Hu Xiao Award and the Second-hand Marketing Society has been released one after another, which has aroused the attention and heated discussion of all parties in the industry.
In order to help marketing practitioners understand the core competence, present situation and future trends of various industries more clearly, we launch a column of digital marketing industry analysis every two weeks, focusing on three sections of China Digital Marketing Ecological Map (202 1 Edition), deeply analyzing the 16th National Congress circuit, interpreting the core competence, present situation and future trends of various industries circuit for you, and helping enterprises to make strategic layout of digital marketing.
The fourth column of digital marketing industry analysis focuses on the "marketing data" in the "Data and Tools" section of Digital Marketing Ecological Map of China (202 1 Edition).
If creativity is the soul of marketing, then data is the blood of marketing and the source of life.
But the big data in our mouth is not that the bigger the data, the better. Data itself is actually a burden, which requires spending money on space storage, continuous maintenance and data mining. So real big data has several basic characteristics:
First, the amount of data is large, which is not the data that ordinary computers can open;
Second, the latitude is rich, and a single dimension cannot be called big data;
Third, the calculation call is fast, and it can be calculated in a short time without running for a year or so;
Fourth, the accuracy is high, and a lot of data are confused when they are collected and random when they are calculated, resulting in low accuracy and losing the meaning of the data;
Fifth, there must be "value".
Today, let's discuss which data are more valuable to us in the marketing field and how we can use them.
There are many types of data we can collect from marketing contacts, which will involve different technologies and channels. (as shown in the figure below)
These data are the data with the simplest technology, the lowest cost and the largest amount of data that enterprises can collect according to the difficulty of collection and the relevance of sales. (as shown in the figure below)
Before using so much complex data, we must have an overall concept and classification of data use rights. For marketing data, the whole industry can be divided into first party, second party and third party data. Another dimension can also be divided into public domain data and private domain data.
The data generated in the process of production, supply and sales belong to the first-party data. Among them, the first-party data sources commonly used in marketing are as follows:
0 1
customer relationship management
That is, using the data generated by the customer relationship management system to record the information of consumers who have purchased products or expressed their purchase intention.
02
SCRM
SCRM can be regarded as an extension of CRM, which is mainly a customer management system based on social systems, such as WeChat Ecology. Many advertisers directly establish CRM on WeChat official account, and consumers can browse historical purchase records, membership points and redeem coupons on WeChat official account.
03
Cytidine 5- diphosphate (short for cytidine 5-diphosphate)
A data platform integrating data collection, data access, unified ID systematization and analysis and mining. The data will be processed twice so that the business parties can use it quickly.
04
User loyalty platform
Membership management system is usually regarded as an advanced version of CRM.
05
Business operation data
Data generated during business operations, such as bank credit card data. With the improvement of our ability to collect and process unstructured data and semi-structured data, we can collect and analyze a large number of conversation data with consumers, such as text and audio conversation records, which can be analyzed after intelligent identification and encryption technology.
Differences between first-party data and private domain data
We can simply understand that private domain data is a subset of first-party data, mainly for "people"-related data. To some extent, the data related to "people" that an enterprise can obtain can be regarded as private domain data, even if it is not generated on its own system, such as data under its own Tik Tok, Little Red Book and e-commerce account. Not all data related to "people" are regarded as private domain data, and we usually only regard the data that enterprises interact with people as private domain data.
Importance of private domain data
0 1
high potential
Private domain data is the data that enterprises interact with consumers. Whether existing consumers or potential consumers, interaction itself represents high activity, which can achieve more transformation and secondary marketing.
02
Combination of public and private domains
The private domain system of media giants can connect with public domain data to realize large-scale acquisition, refined follow-up and secondary transformation;
03
safe
With the advancement of the personal information protection law, the use of public domain data will become more and more cautious. Every private domain system is relatively independent, and private domain data can protect consumers' data from being fully utilized in a certain range.
If the first, second and third-party data are taken as a whole, the first-party data is usually closer to the business of the enterprise, but the quality of the first-party data is very limited, the data width is not enough (quantity), the data thickness is not enough (field retention rate), and the data refresh speed is not enough (update frequency). In order to get better data model results, we usually need to rely on external data.
Second-party data is usually related to your own enterprise, but you have no ability or right to collect it, and you need to entrust or rely on other parties to collect or provide it. The data related to your own enterprise is only provided to your own enterprise.
For example, account operation data in the WeChat system, operation data on e-commerce, and monitoring data collected by third parties.
Data that can be publicly obtained or traded. Marketing-related, such as industry reports, third-party DMP platforms, external data providers, data trading platforms, data trading blockchains, etc.
It embodies the real-time nature of digital marketing. Not everything needs to be presented in real time. The four dynamics in the whole scene are produced to meet the needs of programmatic buying.
Case 1:
Dynamic budget allocation within the group
0 1
background
The purchase of super-large customers is usually concentrated, and enterprises have the problem of multi-brand and multi-target people. In the environment of programmatic buying, it is a challenge to scientifically allocate the purchase quantity to different brands in the same period and to distinguish the priority of people and labels. It is necessary to find a scientific, reasonable and dynamic way to distribute and manage the flow within the group in time.
02
tactics
Using advertising service technology to realize dynamic traffic distribution
Confirm multi-brand strategic relationship
Analyze the difficulty of label acquisition, customer value, label accuracy, etc. , and put forward the priority recommendation. For example, mothers and babies have high customer value, high judgment accuracy and great difficulty in obtaining, so they give priority to displaying the corresponding brands than other labels.
When reach is the goal, the overclocking part will be replaced by other brands.
(e.g. brand strategy)
03
affect
Overall flow efficiency is improved 15-30%.
Case 2: "Tailor-made"
Fine user access, dynamic crowd communication
0 1
background
In the listing stage of a new automobile brand, we hope to expand the cognition of new users and let more core consumers know and be interested in the products.
02
solution
Real-time dynamic optimization level material rotation combining crowds and creative materials in stages.
Real-time delivery redirection of potential interest groups
03
Project results
Case 3:
Dynamic management of content customized for the whole network
0 1
background
As a high-end industrial product, automobile brand has a far-reaching impact on product sales. In the practice of a car company, it is found that the consumers of car companies, especially the younger generation, are not clear about the brand and marketing content: car companies don't know whether their content ideas meet the preferences of target consumers; Car companies don't know whether the content has been accurately delivered to the target consumers.
02
solution
Provide a working system (full data operation+full process intelligence) and two management platforms (social content analysis platform+content management platform) to optimize content management and matching:
Firstly, the content is managed in data to establish the foundation for managing the content;
With the foundation, we will continue to explore and dig, and solve the matching problem between content and customer preferences through insight analysis and prediction analysis;
Finally, the process of content management is intelligent through content generation and recommendation, which is helpful to realize the closed loop of operation.
Based on user information, dynamically generated creative content is displayed on the visit page (official website/Media).
03
result
Through the implementation of the project, car companies have realized multi-touch refined operation, greatly improving the consumer experience and customer loyalty, which is embodied in:
Through social network public opinion insight analysis, real-time monitoring of car brand image, the overall net goodwill increased;
According to different consumers, after the creative marketing scheme of a certain model was realized in official website, the amount of retained funds of this model increased by about 40% through A/B test.
When we talk about digitalization, we usually present and collect all the points in the process in the form of data. In this process, with the opening of internal and external data and intelligent decision-making. Some head advertisers have set up BTD(Brand Trading Deck) to open all the uploading and publishing links according to their own processes. Realize the whole process from planning, ordering and evaluation online. And by using historical data and models, intelligent allocation, decision-making and prediction can be made.
Case 1: AI aided decision-making,
Intelligent allocation of media budget
0 1
background
Customers have a lot of data scattered in the hands of different agents and suppliers, lacking unified management; The process involves many internal and external departments and lacks the ability of authority control; There are many revised versions of historical production scheduling, so it is impossible to make a resumption decision; Lack of supervision and management and transparency; Lack of scientific decision-making basis, relying on personal experience can not meet the increasingly complex media changes.
02
respond
The whole process is online, and the API of each link is connected to realize the unification of the system and the view;
Reorganize the authority management system;
Docking the media inventory system, importing the customer's own historical data and making a landing model;
The automatic distribution of policies is realized by API+Email;
Empirical decision-making and model decision-making coexist to allocate online and offline budgets.
03
result
Completed the digitization of media assets;
Established a traceable and transparent decision-making process mechanism;
Reduce the comprehensive cost of media by 5% (the price itself is already the most competitive in the industry);
The media operation cycle has been shortened by 70%.
Full-process intelligence is the deep mining of data, which maximizes the value of data through algorithms and models. Can play a role in all aspects of closed-loop marketing.
In the stage of strategy optimization, exploratory insight mining can be carried out based on knowledge map; In the early analysis stage, we can predict the marketing objectives by analyzing the past data; In the delivery stage, the most suitable content is pushed according to the intelligent judgment of the crowd, and can be continuously optimized. Intelligent exploration is a relatively new field, and it is also a technology that may affect the whole consulting and marketing industry in the future.
Knowledge map itself is a structured semantic knowledge base, which describes concepts and relationships in the physical world in the form of symbols. The basic unit is the "entity-relationship-entity" triad. There are several use directions in the marketing field. For example, exploratory active recommendation. Our traditional search is passive correlation. For example, if you search for a mobile phone, the results are all kinds of mobile phones or mobile phone cases. Exploratory recommendation is to actively explore the best answer after analyzing the correlation and relationship between various information.
Marketing itself is a process of exploration and innovation. If the machine can give more accurate relevant results through big data, it will bring a revolution to the advertising and consulting industry. Once (service stopped on February 19, 2022), there was a search website based on knowledge mapping technology called Magi (pictured).
Its appearance makes Baidu feel uneasy, and the active exploratory recommendation method re-associates people, things, things and concepts, subverting Baidu's paid ranking and relevance ranking model. In addition to exploratory recommendations, we also see some marketing applications that use knowledge mapping technology, such as discovering consumer needs through social networks and transforming them into new product development. Automatically generate text ideas, automatically analyze consumer intentions and so on.
Case 1:
Using knowledge map to analyze customer needs
The direct application of knowledge map in marketing can help us to tap the potential needs of consumers and better match the corresponding communication and products. For example, the following figure discusses the possibility of consumption and the types of products that may be consumed by using consumers' various contact behaviors.
Case 2: intelligent prediction,
Model algorithm is used to predict and optimize CTR.
Case 3:
Market budget allocation model
Case 4: beautiful customers pass the customs,
Similar model promotes sales transformation
(Beauty customers use Lookalike model to improve the effect)
Case 5:
Automobile customers intelligently optimize the potential population.
0 1
background
Car companies put in a lot of money every day, and in the process of programming, they reached most netizens through different channels. There are many potential intentional people, but it is impossible to accurately identify and reach them again.
02
target
Integrate multi-party data and build a set of ID scoring model, so as to distinguish different potential groups and improve the efficiency of fund retention.
03
way
The first step is to collect all the crowd equipment IDs contacted by advertisements, and screen out the retained (data from the owner's APP and other channels can be used later) as positive samples for training;
The second step is to identify these people's media contact habits through model learning, such as which website they fall in love with, which will change after reading it several times. Brand website behavior, such as how many pages you usually read and how long you stay, has a strong willingness to stay. Match the external interest labels to see which hobbies you prefer;
The third step is to distinguish people, communicate differently, avoid low-quality people and follow up high-quality people with high frequency.
04
result
The efficiency of people with high tendency is 28 times that of the control group.
This part focuses on the ability of enterprises to build their own data platforms. For business and marketing departments, they are generally not from technical background, and there are barriers to the choice of this data platform. We provide a set of classification and selection methods for your reference.
Selection of data platform
For many enterprises, the choice of what kind of data platform to build directly determines the success or failure of the project and even the whole company strategy. We can simply summarize it into three technologies and four modes. These three technologies are DMP, CDP and data center (data lake). Let's make a simple distinction between the three most confusing technology platforms from three angles with the following figure. In addition to the most understandable functional dimension and data dimension, application scenarios are also very important. It affects the timeliness and data accuracy required by this platform, and the gap is very large.
What department is suitable to take the lead?
Make the corresponding platform.
Data lake or data center is more suitable for the overall digital transformation of the company, while DMP and CDP are more suitable for marketing and business departments.
What kind of industry and stage is it suitable for?
Create a corresponding data platform
Understanding the application scenario depends on the characteristics and stages of the customer's industry.
Four mode choices
After choosing three technologies, we need to look at what mode to use when landing. The factors to be considered here are the company's existing technology platform, the company's expectations, the company's technology policy, the company's budget, and the corresponding talent pool.
We provide three construction ideas for your reference:
Challenges faced by data platforms
0 1
ID can't get through, resulting in data island in data center.
ID can't get through, which is one of the biggest challenges facing today's big data platform. There are several reasons, on the one hand, technical reasons. It is difficult to identify heterogeneous data from multiple sources. On the other hand, due to policy reasons, such as the prohibition of private data by the state, the conversion between mobile phone number and device ID is typical. All major platforms and customers also have their own policy data, which can't be casually. There is also a problem of process organization. The cooperation between departments is not good, each department is fragmented, the wheels are repeated, and the data are not unified. There is also a lack of process, and there is no handshake mechanism designed during data collection, which leads to data fragmentation.
02
Return on investment trap, unreasonable expectation
The construction of one-party data platform is the symbol of refined marketing operation. At the beginning, it is impossible to receive returns as quickly as extensive marketing. Instead, we need to build a data platform with the thinking of product managers. It is not only a tool to treat headaches, but also a data barrier in the future. Therefore, it is necessary to have an overall plan and have enough time to do preliminary design and gradually realize its value. Many projects are often sacrificed because of high expectations. Therefore, we should see the inflection point effect of ROI and treat the growth of data platform with reasonable expectations.
03
Talent resources can't keep up.
One-party marketing data platform is an area that both business and technology need to know. Most of Party A's IT departments still play the role of supporters, and there are serious deficiencies in the understanding of business needs and the level and number of personnel. Most outstanding young technicians have been poached by big Internet companies with high salaries, resulting in the embarrassing scene that the system is not maintained and used. Finally, they have to choose SAAS mode or rely on suppliers.
Today, China's marketing data is at a crossroads. Personal privacy protection law challenges data collection, transmission and storage, and all families are exploring the boundaries of policy and technological breakthroughs, such as federal learning and CDP. When the use of external data becomes more cautious, the utilization rate of one party's data will become a barrier.
At the same time, in the future, institutions with government background need to take the lead in formulating new standards that can meet national requirements and promote the development of the data industry in a mature and compliant direction.
Current and previous reports
Production | research group
Finishing | He Yuqing
Editor | Liu Zhaolong
Editor on duty | Wang Linna
Help the digital transformation and upgrading of enterprise marketing
we have
200+ Head Digital Marketing Industry Resources
More than 500 frontline combat experts
10000+ excellent award-winning cases
If there is a demand for marketing digital consulting,
You can contact Ms. Jiang
Tel:13851790129.
E-mail :hx005@hooxiao.com.
Theme planning: theme planning: theme planning: theme planning 丨丨丨丨丨丨丨丨丨丨丨丨丨丨
Interview with new consumers: 丨丨丨丨丨丨丨丨丨丨丨丨丨丨丨丨丨丨丨丨丨丨
Interview brand: 丨丨丨丨丨丨丨丨丨丨丨丨丨丨丨丨丨𰵈.
Want to join more quality communities
Please scan the QR code below to add Xiaoxiao WeChat.
International Brand Watches —— Digital Marketing in 2022
The new issue in May has been launched!