Useful information | User portrait methods and practices in operational data analysis!

In daily business activities, there are various functional divisions: growth, content, activities, and products. Although the specific work and ultimate goals are different, they are actually all centered around "users" and can be said to be all doing " user operations ." Now, with the end of the traffic dividend and the continuous increase in customer acquisition costs, we have entered a stage of refined user operations. At this stage, we have to use a tool - user "portrait" labeling system.

Today we are going to talk about user portraits . The focus of this article: 1. The application of user portraits in data analysis; 2. How to build user portraits.

The concept of User Persona was first proposed by Alan Cooper, the father of interactive design. It is a target user model based on a series of attribute data. Generally, it is a typical user abstracted from the user group by product designers and operators. It is essentially a tool to describe user needs.

Personas are a concrete representation of target users.

Virtual representation of real users

——Alan Cooper, the father of interactive design

But with the development of the Internet, the user profile we are talking about now includes new connotations: a labeled user model abstracted based on information such as user demographic characteristics, web browsing content, online social activities and consumption behavior.

Its core work is to analyze and mine the massive logs and large amounts of data stored on the server and to label users. A "label" is an identifier that represents a certain dimension of a user's characteristics and is mainly used for business operations and data analysis. (As shown in the picture)

Users play a vital role in the development of enterprises. The main applications are:

1. Precision marketing: This is the most familiar way of operation. In the process from extensive to refined operation, the user group is cut into finer granularity, supplemented by SMS, push, email, activities and other means, to drive strategies such as care, recovery, and motivation.

2. User analysis: User portraits are also a necessary supplement to understanding users. In the early stages of a product, product managers learn about users through user surveys and interviews. As the number of product users increases, the effectiveness of the survey decreases. At this time, user portraits can be used to supplement the research. The directions include the characteristics of new users, whether the attributes of core users have changed, etc.

3. Data application: User portraits are the basis of many data products, such as the well-known recommendation system and advertising system. Advertisements are placed based on a series of demographic-related tags, such as gender, age, education, interest preferences, mobile phones, etc.

4. Data analysis: User portraits can be understood as data warehouses at the business level, and various labels are natural elements of multi-dimensional analysis. The data query platform will be connected with these data and ultimately assist in business decision-making.

User portraits are generally divided into multiple category modules according to business attributes. In addition to common demographics and social attributes, there are also user consumption portraits, user behavior portraits, user interest portraits, etc.

Demographic attributes and behavioral characteristics are included in most Internet companies when creating user portraits: demographic attributes mainly refer to the user's age, gender, province and city, education level, marital status, fertility status, industry and occupation, etc. Behavioral characteristics mainly include indicators such as activity and loyalty.

In addition to the more common features mentioned above, the content of user portraits is not completely fixed, and the features that are focused on vary depending on the industry and product.

① Content-based media or reading websites, search engines, or general navigation websites often extract user interest characteristics in browsing content, such as sports, entertainment, food, finance, travel, real estate, cars, etc.

② The user portrait of the social networking site will also extract the user's social network, from which we can find closely related user groups and star nodes that act as opinion leaders in the community.

③The user portrait of the e-commerce shopping website generally extracts indicators such as the user’s online shopping interest and consumption capacity. Online shopping interests mainly refer to users’ category preferences when shopping online, such as clothing, luggage, household items, mother and baby items, cleaning and care items, and food items. Consumption capacity refers to the user's purchasing power. If it is done in detail enough, the user's actual consumption level can be distinguished from the psychological consumption level in each category, and characteristic dimensions can be established separately.

④ In the financial field , there will also be risk profiling, including credit reporting, default, money laundering, repayment ability, insurance blacklist, etc.

In addition, the user's environmental attributes can also be added, such as the current time, LBS characteristics of the visited location, local weather, holidays, etc. Of course, for specific websites or apps, there are definitely user dimensions that require special attention, and these dimensions need to be further refined so that users can be provided with more accurate personalized services and content.

There are many methods for creating user portraits in the industry, such as Alen Cooper’s “Seven-Step Persona Method”, Lene Nielsen’s “Ten-Step Persona Method”, etc. These are very good and very professional user portrait methods, which are worthy of our reference and learning.

In fact, after we understand these methods, we will find that these methods can be divided into three steps in terms of process: obtaining and studying user information , segmenting user groups , and establishing and enriching user portraits . Among these three steps, the most important difference lies in the acquisition and analysis of user information. From this dimension, there are mainly three methods:

Simply put, qualitative analysis is to understand and analyze, while quantitative analysis is to verify. Generally speaking, quantitative analysis is more expensive and relatively more professional, while qualitative research is relatively cost-effective. Therefore, the method of creating user portraits is not fixed, but needs to be determined according to the needs, time and cost of the actual project. There is no strictly most professional or scientific method for creating user portraits, but there is the method that best suits the needs of the team and project.

A good user portrait understands the user's decision-making and takes into account business scenarios and business forms. Here we introduce a simple method to build user portraits.

1. Data Collection

The purpose of building a user portrait is to restore user information, so the data comes from all user-related data. User data is divided into two categories: static data and dynamic data.

Static data: There are five dimensions of users’ demographic attributes, business attributes, consumption characteristics, lifestyle, and CRM. There are many ways to obtain it. Data mining is the most common and relatively accurate method. If the data is limited, it needs to be supplemented by a combination of qualitative and quantitative methods. Qualitative methods such as focus groups, in-depth user interviews, diary method, laddering method, transmission method, etc. mainly use open-ended questions to penetrate into the real psychological needs of users and visualize user characteristics; quantitative methods are more often conducted through quantitative questionnaire surveys. The key lies in the modeling and analysis of quantitative data in the later stage. The purpose is to verify qualitative hypotheses through closed questions on the one hand, and to obtain the user distribution pattern in the market on the other hand.

Dynamic data: User behavior information that is constantly changing. For example, a user opens a web page and buys a cup; this is also the same as the user taking the dog for a walk in the evening, withdrawing money during the day, yawning, etc., which are all user behaviors. With the development of the Internet, all kinds of dynamic behavioral data can be recorded.

2. Target analysis

The goal of user profiling is to analyze user behavior and ultimately label each user and assign a weight to the label. 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.

3. Data Modeling

An event model includes three elements: time, place, and people. Each user behavior is essentially a random event, which can be described in detail as: what user did what at what time and where.

① User: The key lies in user identification. The purpose of user identification is to distinguish users and single point positioning.

②Time: Time includes two important information, timestamp and time length. Timestamp is used to identify the time point of user behavior; time length is used to identify the time a user stays on a certain page.

③Location: user contact point, Touch Point. For every user touchpoint. Potentially contains two layers of information: URL and content. URL: Each link (page/screen) points to an Internet page address, or a specific page for a product. It can be a page of an e-commerce website on a PC, a function page of applications such as Weibo and WeChat on a mobile phone, or a specific screen of a product application. For example, the product page of Great Wall Wine, the WeChat subscription account page, and the level-clearing page of a certain game.

④Content: The content of each URL (page/screen). It can be relevant information about a single product: category, brand, description, attributes, website information, etc. For example, red wine, Great Wall, dry red wine, for each Internet contact point, the URL determines the weight; the content determines the label.

⑤Things: User behavior types. For e-commerce, the typical behaviors are as follows: browsing, adding to shopping cart, searching, commenting, purchasing, clicking likes, collecting, etc.

Based on the above analysis, the data model of user portrait can be summarized as the following formula: user ID + time + behavior type + contact point (URL + content), which refers to what a user did at what time and place. So it will be labeled.

The weight of user tags may decay with time, so time is defined as the decay factor r. The behavior type and URL determine the weight, and the content determines the tag, which is further converted into the formula: tag weight = decay factor × behavior weight × URL sub-weight.

1. Don’t use typical users as user profiles

Typical users cannot be used as user portraits. In the WeChat Life White Paper every year, WeChat officials will publish a typical user’s day: getting up at 7 a.m. on weekdays to check Moments, reading articles on the way out at 7:45... Many users said that this is exactly themselves! However, many people also complained: I am also a heavy user of WeChat, but why does this typical day not match me at all?

Why are there such contradictory feedbacks? It turns out that these people confused the concepts of "typical users" and "user portraits." Because the above descriptions of the characteristics of typical users are just abstract user characteristics and combined together. In fact, typical users are fictional and do not really exist. User portraits represent users in the form of labels, and every real user has a corresponding user portrait.

2. Don’t simply understand user portraits as consisting of user tags

This is also a misconception that more than 50% of people may have, that is, simply understanding user portraits as consisting of user labels. User tags are used to summarize user characteristics, such as name, gender, occupation, income, cat ownership, preference for American TV series, etc. On the surface, there seems to be nothing wrong with these labels, but in fact, the labels that make up the user portrait must be combined with the business/product.

To give an exaggerated example, Haidilao wanted to create a user portrait, and finally listed user labels such as Xiao Ming is a college student, tall, rich and handsome, an only child, from Sichuan, who likes to play games and watch anime, etc. In fact, for Haidilao, it really doesn’t matter whether the user is handsome or likes to play games.

3. Failure to establish truly effective user portrait labels

If you can establish truly effective user portrait labels, you can be considered to have a correct understanding and thus improve operational results. This involves the biggest difficulty in building user portraits.

For example, if a knowledge payment team wants to sell courses, then the core demand of establishing user portraits is to increase the number of course purchases. If we can understand users' willingness to purchase courses through user portraits and then adopt corresponding operational strategies, efficiency will be greatly improved. The willingness to purchase courses is the label that we most need to put in the user portrait.

For example, after we established user portraits, we calculated that user A’s willingness to purchase the course is 40%, while user B’s willingness to purchase the course is 90%. In order to further increase the purchase volume, we will issue coupons to users (A) with a purchase intention of 40%. If we don’t establish such a user portrait tag, we will issue the same coupon to A and B. Class B users do not need to be incentivized with coupons, and issuing them will increase costs a lot. This is how e-commerce companies use user portrait tags to achieve big data price discrimination.

1. We have entered a stage of refined user operations. At this stage, we have to use a tool - user "portrait" label system;

2. User portrait is a labeled user model abstracted from information such as user demographic characteristics, web browsing content, online social activities and consumption behavior;

3. Users play a vital role in the development of enterprises. The main applications are: precision marketing, user analysis, data application, and data analysis;

4. The content of user portraits is not completely fixed, and the features they focus on vary depending on the industry and product.

5. A good user portrait understands the user's decision-making and takes into account business scenarios and business forms.

Source: DataHunter