Whoever wins the user wins the world. User portrait has become the key to precision marketing

1. The Origin and Development of User Portraits

Since Sir Tim Berners-Lee invented the World Wide Web (WWW) in 1991, after more than 20 years of development, massive DT-level data has emerged, making it impossible for any individual or enterprise to digest and absorb independently. The Internet has entered a new stage: the era of big data .

A few years ago, this concept was once over-hyped. In recent years, people have gradually calmed down and focused on how to truly utilize big data, tap into potential business value, and how to truly apply big data technology in enterprises.

In the discussion of big data applications, personalized marketing has gradually become a very important landing point in its commercial applications. Compared with less efficient methods such as traditional offline questionnaires, big data enables companies for the first time to conveniently obtain more extensive feedback information from users through the Internet, providing sufficient data foundation for further accurate and rapid analysis of important business information such as user consumption habits.

This will inevitably bring about a series of dual changes and shaping in corporate marketing and consumer behavior. The biggest change is that all consumer behaviors seem to be "visualized" in front of companies.

With the further in-depth research and application of big data technology, enterprises are increasingly focusing on how to use big data for precision marketing services and further tap into the potential commercial value of customers.

Thus, the concept of "user portrait" came into being.

The purpose of user profiling is to abstract the entire picture of a user's information as comprehensively as possible, providing sufficient data foundation for further accurate and rapid analysis of important information such as user behavior habits and consumption habits.

It is an in-depth exploration of people. In addition to objective attributes such as basic demographic information and geographic location, attributes such as interests and hobbies are highly subjective, change rapidly, and are difficult to count.

However, in many commercial application scenarios, advertisers or demanders want precisely these highly subjective attributes, including a person’s values ​​and personality, etc.

For example, among automobile customers, environmentally friendly electric car brands most want to reach those customers who are environmentally conscious and like small-displacement vehicles. This actually involves people’s values ​​and car-buying preferences at the same time.

These user attributes are actually "user labels". To find these labels, it is impossible to simply discover them through direct analysis of user behavior. A deeper exploration of personality is required.

In short, through "user tags", companies can put aside personal preferences during the advertising process and focus on the motivations and behaviors of target users to carry out marketing activities . Because designing advertisements for specific user portraits is far more effective than designing for something fictional in your mind, and it is also easier.

2. What is the relationship between user portraits and tags?

The essence of user portrait is the “labeled” user behavior characteristics.

The core of user portrait production is to "label" users, such as age, gender, region, user preferences, etc. Each label is usually an artificially defined feature identifier, using highly refined features to describe a type of person.

Therefore, user portraits can provide companies with sufficient information foundation, helping them quickly find accurate user groups and more extensive feedback information such as user needs.

At the same time, after abstracting the typical user model from the real user behavior, the enterprise can collect and analyze the consumers' main behaviors such as social attributes, consumption behaviors and living habits, and fully describe the target user characteristics of products and services. This can provide effective information for all user-related decision-making processes in the enterprise and guide the enterprise's product and service research and development and marketing.

User tags usually have two important characteristics: semantics and short text.

Semanticization allows us to easily understand the meaning of each tag, which also makes the user portrait model meaningful and can better meet business needs.

Short text, that is, each label usually only represents one meaning, and the label itself does not need to do too much preprocessing such as text analysis, which makes it convenient to use machines to extract standardized information.

At the same time, labeling rules are set by humans, so they can quickly read the meaning of a large number of labels; while machines efficiently perform label extraction and aggregation analysis. Therefore, the user tags in the user portrait show us a simple and concise method to describe user information.

Generally, there are two types of tags: one that can be structured and one that cannot be structured.

Structured labels have clear divisions and can be exhaustive, such as gender, age, and occupation.

However, once this label is affixed, it often has a time limit. Medium- and long-term user tags are relatively stable and are characteristics of users that change less frequently, also known as static tags; medium- and short-term user tags change slightly faster, such as interest tags, also known as dynamic tags.

(Structured medium and long-term label illustration)

(Structured short- and medium-term labels)

Unstructured tags have no clear division, are not fixed, and are huge in number. The worst thing is that they change very quickly, such as App tags and personalized tags. It is precisely because of the existence of this type of label that the difficulty of depicting accurate user portraits has increased exponentially.

(Unstructured label illustration)

After we finally label the user portraits, we must also label the weight of each label. For example, red wine 0.8, Nike 0.6. Tags represent content, users’ interests, preferences, needs, etc. Weight represents the index of user’s interest and preference, and may also represent the user’s demand, which can be simply understood as credibility and probability.

Generally speaking, label weight = time (when) x location (where) x behavior (why)

Every user behavior can be described in detail as: what user, at what time, at what place, and what happened.

For example, user A bought an iPhone 7 on Apple’s official website yesterday, and the user label reflected was “Apple fan 0.9”, while user B collected an iPhone 7 on Tmall last month, and the label reflected was “Apple fan 0.498”.

These different user tags and their corresponding weights will play an important guiding role in subsequent marketing decisions.

At the same time, user labels and weights in user portraits also provide a convenient and efficient way for computers to process big data, allowing computers to programmatically process information directly related to people, and even through certain algorithms and models, to "understand" people to a certain extent.

When computers have the ability to understand user portraits, the efficiency and accuracy of applications such as personalized recommendations and advertising will be further improved.

3. Why is user portrait so important in today’s advertising and marketing?

As mentioned earlier, the emergence of user portraits actually stems from their application in the business field, that is, the company's need for user cognition. In marketing activities, what advertisers fear most is wasting money, and most of the reasons for wasting money are inaccurate user targeting.

In the process of marketing decision-making, companies are concerned about two issues:

1) How to make a product that customers like more?

2) How to sell products to the right people?

Solving these two problems requires insight into user needs, so decision makers inevitably have to consider two types of people:

Existing Customers: Who are my existing users? Why buy my product? What are their preferences? Which users are the most valuable?

Potential Customers: Where are my potential customers? What do they like? Which channels can find them? What is the customer acquisition cost?

In the field of precision marketing, user portraits are indispensable, which means that in the era of mobile marketing, whoever wins the users wins the world.

Today's marketing wars are no longer just about creativity, and companies are thinking more and more deeply from the user's perspective. Without talking about precise delivery, data analysis and user profiling, companies would be embarrassed to say that they are still doing marketing.

Only by finding target users more accurately based on product characteristics, delivering content on channels preferred by users, and promoting purchasing behavior through timely interactions can we achieve accurate customer acquisition.

The user portrait formed based on big data technology has opened up a whole new picture for the marketing industry. It gives marketers the opportunity for the first time to contact and analyze users from a higher-level dimension, and allows them to see for the first time in user portraits a treasure trove of information that they have never been exposed to and are extremely excited about.

From the user portrait, advertisers (stores, merchants, and any demand side) can select and decide the users they want to reach and how to reach them through the labels and weights in it. Internet companies such as BAT rely on their natural advantages and a large part of their advertisements reach users in this way.

360 has been immersed in the field of advertising and marketing for many years, and advertising revenue has always been the bulk of the company's revenue. Everyone from the boss to the employees attaches great importance to it, so it has made early arrangements in the field of corporate marketing driven by new technologies such as user portraits, personalized recommendations, and big sentence algorithms.

360's product matrix covers a total of 500 million PC users and 800 million mobile users, which makes its data reserves completely sufficient. At the same time, the diversity of its data (its own, customers' and third-party sources), cross-screen level (PC, mobile phone and TV) and technical capabilities of data processing have reached the top domestic level.

360 Shangyi is the first intelligent big data marketing decision-making platform launched by 360 Company. It can accurately locate the target audience, analyze the natural attributes and online behavior of the audience, analyze the market status of the industry and brands, and truly provide advertisers with real, accurate and reliable decision-making support for their placement and marketing.

For example, from the marketing case of Douyu TV, 360 Shangyi found that most users visiting Douyu are male students aged 19-24, and they are very interested in digital products and food. This precise positioning of users is very important for advertisers.

If advertisers want to deliver more precise advertising with a higher ROI, they can start from 360 Shangyi’s analysis and deliver advertising to people in more differentiated interest categories. If advertisers want to expand their influence and allow more people to have the opportunity to know their brand, they can consider placing ads in categories with higher sample coverage.

At the same time, 360 DMP provides flexible combination of screening conditions to achieve population segmentation management, from user behavior trajectory to interest preferences and other dimensions, to freely combine and circle the target population for delivery.

When placing display, search, brand, and app download ads, it can quickly and accurately target a certain type of population, thereby helping advertisers achieve higher ROI and effectively reduce marketing costs.

Take the Huajiao Live App invested by Zhou Hongyi as an example. In the era of universal live streaming , the market competition among various live streaming products is fierce. Therefore, Huajiao Live hopes to reduce the activation cost of new customers while acquiring more new customers.

360 DMP found that people who are more interested in live streaming often visit similar live streaming websites, search for related keywords , and download industry-related apps. Therefore, people who have these behaviors are potential users of Huajiao Live.

Relying on 360's own precise user portraits, DMP helped Huajiao Live perfectly match those who might be interested in live streaming apps. After it placed advertisements, the activation rate of the Huajiao App increased by nearly 3 times!

4. The difficulty of accurate user portrait is far beyond imagination

However, opportunities and challenges always coexist. An accurate user portrait can bring considerable advertising revenue, but it is likely to encounter the following difficulties:

1) User entity identification: Family members often share computers or iPads, and it is not as easy as you think to find the person behind the computer or phone;

2) Definition of user tags: Individual tags are useless by themselves. They must be linked together to form an overall user profile and combined with the advertiser’s actual business scenarios to be meaningful.

3) Data processing capabilities: The processing of massive amounts of labels, especially the processing of unstructured labels mentioned above, is the time to test big data technology;

4) User portrait mining and modeling: The accuracy of user tags needs to be continuously optimized. It is necessary to combine natural language processing technology, machine learning, deep learning and other methods to continuously optimize the effect, and combine it with the specific needs of advertisers;

5) Verification of user tags: Some tags, such as age, can be verified for accuracy using standard data sets. However, some tags, such as interests and user loyalty, do not have factual standards and can only be verified for effectiveness in specific businesses.

Ultimately, labels are just terms, technology is just a form, and user portraits are ultimately meant to serve companies’ marketing decisions about their customers.

In the psychedelic fog of the market, user portraits are like a bridge that connects the information and cognitive gap between enterprises and users. Through big data technology, users' complex consumption behaviors and elusive psychological states are presented to enterprises (demand side) in a more rational way, namely user portraits, so that enterprises can truly make rational and well-founded marketing decisions and win the battle thousands of miles away.

Reference Documents:

"Bringing Big Data to Ground: The Application of User Portraits in 360 Business Data Department"

"Unleashing the Value of Big Data: How to Create Accurate User Portraits?" 》

Reference author:

Xiang Yuan, an algorithm engineer in 360 Business Data Department, holds a master's degree in pattern recognition from the Beijing University of Aeronautics and Astronautics. He is engaged in the mining of DMP user portraits and the development of models and algorithms for advertising targeting.

Du Ranran, an algorithm engineer in the 360 ​​Business Data Department, has worked in Alibaba and 360, engaged in natural language processing-related work such as search algorithms, data mining, and advertising algorithms.

About 360 Business Data:

360 Business Data Department focuses on in-depth mining and analysis of 360's own massive data. While protecting personal privacy and data security, it analyzes user needs and preferences in multiple dimensions, uses data mining and artificial intelligence technologies, and scenario-based applications to comprehensively enhance business value. It has formed a data marketing product system including 360 Shangyi, 360 DMP and 360 Analysis. 360 Shangyi uses massive data to gain insights into crowd portraits and brand status, providing support for marketing decisions; 360 DMP integrates and manages data, accurately identifies target groups, and improves conversion effects; 360 analysis supports promotion effect evaluation and traffic analysis, and optimizes delivery in real time. This big data product system, combined with the 360 ​​Dianjing effectiveness platform, jointly provides advertisers with a closed-loop service of big data precision marketing.

Mobile application product promotion service: APP promotion service Qinggua Media advertising

This article was compiled and published by @刘胖胖(Qinggua Media). Please indicate the author information and source when reprinting! Site Map