1. Introduction to Toutiao

Industry Overview

According to iResearch data, as of December 2018, the number of independent mobile Internet devices reached 1.347 billion, a year-on-year increase of 12.8%. The number of independent devices in the news and information industry reached 777 million, a year-on-year increase of 18%, and the industry penetration rate reached 57.7%.

Market conditions

With the full arrival of the mobile Internet era, the service threshold and cost of obtaining information through news clients have been greatly reduced, making it an important channel for us to obtain news tips first and also the place where news spreads the fastest.

Judging from iResearch's mobile data in 2018, Toutiao has been ranked second on the news and information list for many years, relying on its algorithm advantages and personalized and accurate recommendations. As of December 2018, the number of monthly independent devices reached 248 million, occupying 32% of the market.

Product Background

1. Product Introduction

Toutiao is a recommendation engine product based on data mining, providing users with a precise and personalized mobile information platform to achieve accurate connection between content and users.

2. Product Features

Based on personalized recommendation engine technology, personalized recommendations are made based on multiple dimensions such as each user's interests and location. The recommended content includes not only news in a narrow sense, but also music, movies, games, shopping and other information.

3. Product Development Process

User attributes

1. User Profile

Gender: Toutiao’s users are mainly male, accounting for 62% and female users for 38%, which is basically in line with the current market situation of news and information apps.

Age: Users are mainly under 35 years old, accounting for 79%. Among them, users under the age of 30 account for 53%, and users as a whole tend to be younger.

2. User usage scenarios

1) When commuting to get off work by public transportation

2) During lunch break

3) Relaxation time before bed

4) Killing time when you have nothing to do or are bored

As the pace of life quickens, people's time becomes increasingly fragmented.

Based on data mining and personalized push notifications, Toutiao maximizes the use of users' fragmented time, achieves accurate connection between content and users, reduces users' time in screening information, and facilitates users' browsing and reading.

2. What is the user life cycle?

To put it simply, the user life cycle is the process from when a user starts to come into contact with a product to when they leave the product. The user life cycle can be divided into five stages: introduction, growth, maturity, dormancy, and churn.

Customer acquisition area: corresponds to the introduction period, the corresponding user behavior is to become a user, and the core task of operation is to attract new users and promote the activity of new users.

Appreciation zone: corresponds to the growth stage and mature stage. The corresponding user behavior is to use the product, be active in the product, establish contact with the product, and continue to stay in the product. The core task of operation is to promote user activity, payment/conversion, and create retention.

Retention area: corresponds to the dormant period and the churn period. The corresponding user behavior is to leave the product and stop using the product. The core task of the operation is to comfort or win back the silent churn users.

3. Build a user life cycle model

As the growth of mobile Internet users approaches saturation, the cost of acquiring new users increases, making the retention of old users particularly important.

Dividing users according to their life cycle can help us understand the needs of users in different life cycles, develop more targeted operational strategies, and thus reduce user churn.

The user life cycle model of Toutiao is built from three aspects:

1. Sort out business logic

2. Find the key features that affect user retention/consumption

3. Define user behavior at each stage

The business logic of Toutiao is as follows:

Based on the above analysis, the basic logic of users using the product is:

1) Whether you can search for the information you want (find information)

2) Whether the information content meets personal preferences (consumption information)

3) An outlet for expressing emotions after reading the news (expressing emotions)

From an operational perspective, the process of users using the product is a cost paid by the users, including time cost, physical cost, mental cost, etc. The higher the cost paid by the users, the stronger their stickiness to the product.

For users, they also need to obtain perceived value while paying the cost, such as the value of the product itself - consumption information, the value of additional services - [wallet] function, the brand value - comprehensive information content, etc. The stronger the user's perceived value, the more likely they are to be active and retain.

Then, based on the user's usage logic, "finding information" is the biggest cost paid by the user, and "consuming information" and "expressing emotions" belong to the perceived value obtained by the user.

Since “expressing emotions” is a derivative value based on “consuming information”, we judge the core key function that may affect user activity/retention: information recommendation.

Define user behavior at each stage

At this point, the user life cycle model of Toutiao has been completed.

From the above analysis, we can conclude that as a traffic-based product, Toutiao can improve user activity and retention by focusing on the number of user logins, product usage time, and function usage guidance.

Let’s use some incentive actions that occur in the product to support the model building content.

The explicit action that occurs

1) Send push notifications when the content of the account you follow is updated, which encourages users to open the APP and increase the number of logins.

2) Through reading behavior, users are given public welfare funds, which can be used for public welfare projects, giving users reading value, satisfying their vanity, and motivating them to use the product for a long time, thereby increasing usage time.

3) When the APP is opened, small dots will appear on the information update page (representing the number of content updates), guiding users to refresh the page to browse the latest information and increase user usage time.

Hidden actions that occur

1) From the perspective of product version iteration, optimizing network loading speed, increasing APP startup speed, optimizing components and other iterative content, as well as maintaining update speed are all ways to increase user usage time and the number of times users use the product by improving user experience.

2) From the perspective of product function analysis, the product has built-in additional functions such as [My Wallet] to meet users' daily life needs and increase user activity.

3) From the perspective of operational methods, users are activated to log in through the continuous entry of celebrities and big Vs, the launch of holiday activities (such as the Spring Festival Chinese New Year event), etc.

4. Design an early warning mechanism to extend the user life cycle

There are two ways to extend the user life cycle. One is to prevent user churn and keep users in the product. The other is to recall lost users. The main discussion here is about preventing loss and designing a loss warning mechanism.

1. Define churned users

Defining lost users can help us make more targeted churn warnings.

As an information product, Toutiao’s core function is reading information. However, users must log in whenever they want to read. User churn can be determined by analyzing their login behavior.

After selecting login behavior as the key churn action for lost users, you can consider the time when the action occurs.

As a high-frequency traffic product, users who have not opened the APP for 15/30 days (internal data can be referred to this time) can be considered to have lost the user. Therefore, we use 15/30 days of no login behavior to determine whether the user has lost the user.

2. Analyze the signs of user churn

User churn behavior is a long-term and continuous behavior. Once signs of churn appear, the early warning mechanism should be triggered immediately. Analyzing the signs of loss can start from the following directions:

Find the reasons for loss from external data Apple store user reviews

To summarize, the reasons why users may churn are:

1) Too many ads

2) Poor content quality

3) Not smooth to use

4) Invasion of user privacy

Internal data can reveal deeper and more convincing sources of loss. Here are a few data dimensions for illustration.

For example, before users churn, they usually have some similar behaviors.

The reasons for churn are determined by segmenting churn behaviors, including visit frequency, average reading time, number of article readings, content opening rate, usage rate of functions such as likes/shares/comments/favorites, and bounce rate.

For example, the distribution of user channel sources, the drainage effect and quality of different channels are different. By analyzing lost users, we can also make a preliminary judgment on the quality of the channel.

For example, the types of lost users, including gender, age, city, interests, occupation, income, etc., need to be included in the data analysis as a basis for judgment.

For example, when users churn, does the product take some actions, such as an increase in churn rate after a version update?

Or some operational measures were triggered, such as a poor experience when launching an activity. These product changes are also key factors in user loss.

For example, from the perspective of product service features, failure to update hot topics in a timely manner and poor content quality will also result in user loss.

Based on the above reasons for loss, we can judge that the signs before user loss are:

1) The number of logins/usage time has decreased over a period of time

2) Not using the product within X days after registration

3. Design an early warning mechanism

Based on the above-mentioned user behavior signs before churn, an early warning mechanism can be triggered. The setting of the early warning mechanism can be manual or automated, and is mainly designed based on the size of the product and the stage of the product life cycle.

Toutiao is currently in a growth stage. Its business model is becoming stable and it has a large-scale user base. It is estimated that it has formed an automated early warning mechanism internally.

4. Complete user onboarding

That is to say, the formulation of strategies to intervene in user churn. This link is very important and requires reference to internal data, so it will not be discussed separately.

Intervention channels:

1) Push

2) Internal message

3) SMS

4) Telephone

5) Email

Since Toutiao belongs to the news and information industry, there is a certain timeliness effect. It is judged that the most commonly used trigger channels are push and in-site notifications.

V. Conclusion

The actual operation of the user life cycle is more complicated. We need to use various means, which is often referred to as refined operation, to achieve the maximum commercial value of users during their life cycle.

Author: Guo Xiaoqian

Source: Guo Xiaoqian