Practical case study: How to apply user portraits?

This article has created a unique 5-step practical method for user operation. Each step is closely linked to the next. Starting from the actual business scenario, it builds a corresponding user model, outputs the target user group and portrait, and then formulates precise marketing strategies to improve the user's life cycle value.

How to use user portraits?

Before raising this question, I studied many articles. The research direction of these articles is more interested in how to give users a 360-degree portrait, using common character image cards to describe the characteristics of a certain part of the platform's users, such as:

• Age: 28, Gender: Female
• Position: Urban white-collar worker
• Marital status: Married
• Personality: A homebody who likes online shopping, beautiful things, literary and artistic, and enjoys life
• Zodiac sign: Scorpio
• Address: xx Community, Beijing

This is a typical user portrait, and it is also a method that many companies are keen on using for user portrait analysis. This kind of character portrait is very helpful in user research and can help product managers find the right direction for the product. However, at the user operation level, if you simply label and profile users, it will feel like it has no application.

Let's take a few business scenarios as examples:

I am an e-commerce platform that sells 100 categories of products. What products should I accurately recommend to users?

I want to promote coupons to users, hoping to increase the ROI of the coupons. To what kind of users should I promote the coupons?

The platform is experiencing a serious loss of users. We hope to analyze which users have been lost and how to retain them?

If you want to use user portraits in the above three business scenarios, how would you apply them?

Obviously, simple user analysis cards cannot support our operational work at all. An important methodology for user analysis is to start from the business scenario, find the precise user group, conduct targeted analysis portraits, and then apply them to actual operational activities.

Next, we will focus on the 5-step method of user operation:

1. Clarify the business scenario

The reason why business scenarios are put first is because business scenarios are the top priority of user operations. Many user operations often fall into a misunderstanding that they first have user portraits. Users are then grouped based on portrait tags and targeted activities are pushed. This operational approach is often marketing for the sake of marketing, but it is impossible to talk about what business problems are ultimately solved.

For example, a certain platform combined several labels to filter out a certain group of people. The labels included: gender, age, most recent purchase time, average order value, etc., and finally launched a full-discount activity. After the marketing is over, we can only analyze the success or failure of an activity, but it lacks practical business support for the overall user operation.

The significance of clarifying business scenarios is that only when we have specific business scenarios can we have specific target groups for marketing, and then formulate more specific marketing strategies based on the portrait characteristics of the groups.

Let’s take the third business scenario as an example. The monthly rolling churn rate of platform users is 30%. We hope to reduce the churn rate to 15% through targeted user operations.

The next task is to find out which users have churned?

Some students may say that this is easy to do. We define platform users who have not returned to purchase for three months as churn. Then we separate these users and push a repeat purchase coupon to them. Then we analyze the redemption effect of this coupon. If the redemption is good, the user churn rate will be reduced.

This is the simplest but least effective way to operate users. I believe many people do this when they first start to operate users.

On the one hand, due to the lack of support from user models, it is impossible to predict churn more accurately. We can only carry out some recall activities based on lost users, but after users actually churn, the recall rate is often difficult to reach 1%. On the other hand, due to the lack of support from portraits and failure to understand the characteristics of the lost user group, it is difficult to impress the lost users by arbitrarily launching some large-scale activities.

2. User Modeling

User modeling is an essential skill for user operations. It is like a technical person who cannot write code and does not understand a coding language, how can he work in this industry?

What kind of user model do we need to build for this business scenario?

We have three approaches:

1. Filter known churned users and build models, analyze churn rules through decision tree models, predict the most likely churned users through churn rules, and filter out these users for marketing.
2. We screen known churned users for modeling, establish a churn scoring mechanism through a neural network model, conduct churn scoring for all high-value users on the platform, and screen out high-value users with higher scores for more precise marketing.
3. Build a user life cycle model, predict user churn nodes through the user life cycle, and conduct timely intervention marketing when users are about to churn to reduce user churn rate.

We adopt the third modeling marketing method and build the following model:

In the model, we can understand the data trend of the entire user base based on the arrows and import all the data of the platform users into the model for analysis.

The Cox survival analysis algorithm is used here, which can help us analyze the survival time of users and obtain the life cycle distribution of all users on the platform. Based on the life cycle, we can obtain the user's churn boundary value and thus determine the user's overall churn node.

3. Identify the target user group

After completing user modeling, we can connect the platform user data with the big data model, and use big data analysis methods to output the user groups we want.

Back to our business scenario, our business goal is to reduce the user churn rate, that is, from 30% to 15%.

So who should our target marketing user group be?

If we simply screen lost users and try to increase the return rate by regaining these users, we know that marketing is very difficult. Generally, lost users tend to uninstall the APP directly or no longer use our products. Once the APP is uninstalled, the reach rate is very low, and the method of recalling them through SMS is often not worth the effort.

Then we turn our attention to users who are on the verge of churn. According to the model prediction, these users have a higher tendency to churn than active users on the platform, but they have not yet churned. At this time, if timely intervention marketing is adopted, the possibility of users staying is very high.

4. Analyze User Portraits

After the target marketing user group is identified, the next step is to analyze the user group portrait to understand the user behavior characteristics, so as to develop a more targeted marketing strategy.

How to analyze user portraits?

The analysis of user portraits here is not comprehensive, but should be based on actual business scenarios. Which portraits are more helpful in formulating marketing strategies? We only need to analyze this part of the portrait information. Let’s first look at the actual portrait of this user group through the model.

Through the model output, we can get the following portrait characteristics of the target marketing population:

1. Female users aged 21-40
2. The average consumption cycle is about 16-30 days
3. The average spending per customer is low, with an average spending of less than 30 yuan.
4. Life cycle is in the rising stage

Based on the above portrait features, what marketing information can we obtain?

1. Gender and age determine our copywriting strategy and communication tone.
2. The number of days between repeated consumption determines the timing of marketing.
3. The life cycle determines different marketing directions.
4. The order volume will determine the intensity of the campaign we promote.

This is the practical application of user portraits in marketing. Every portrait information can be used in marketing strategies. During the modeling process, we need to obtain portrait information that has practical business significance and use it to guide the formulation of subsequent marketing strategies.

5. Develop a marketing strategy

After we get the user portrait, we can carry out more precise and targeted marketing activities based on the portrait information. Next, let's sort out the marketing strategy of this actual combat:

• The target user group is in the rising stage and new customer stage of their life cycle. Their characteristics are that consumption habits have not been formed and stimulating activities need to be planned to guide consumption.
• The consumption characteristics of the target user group belong to the low-order consumer group. The threshold for participating in the activity should not be too high. The activity can have the secondary goal of appropriately increasing the order.
• In terms of portrait characteristics, the consumer group is young women between the ages of 20 and 40, and the communication strategy needs to use a more emotional approach.

Based on the above marketing strategies, we formulate execution strategies for on-site activities and push outreach. I chose a cake in conjunction with Christmas, with the theme: spend Christmas with your beloved one, targeting the group of young women who love romance.

The activity is to redeem this cake at half price for purchases over RMB 35, so as to increase customer orders and stimulate consumption at half price. Screen the target user group, and further screen the members who have not returned for consumption within 16-30 days to send SMS reminders to improve SMS ROI.

Above, we have explained in detail the practical experience of user operation and the 5-step practical method of user operation. Each step is closely linked to the actual business scenario. We build the corresponding user model and output the target user group and portrait, and then formulate precise marketing strategies for marketing to enhance the user's life cycle value.

Author: Zhao Wenbiao

Source: User Operation Observation