This article introduces the three major principles of large-scale user operation systems - data-driven, refined operations, and automation (or productization), as well as the three subsystems in the user operation system - user life cycle management system, user stratification operation system, and user behavior incentive system.

In our actual product work, we often use various operation systems, such as user life cycle management, user stratification, points system, task system, membership system, medal system, etc.

So what is the internal connection between these? How can we control different operating systems and use them in a coordinated manner?

This starts with user operations.

A good product manager must understand users. So what does it mean to understand users?

From a micro perspective, to understand user psychology, one must have empathy and understand the motivations and causes of user behavior in specific scenarios; from a macro perspective, when the user scale of a product reaches a certain level, users begin to be stratified or grouped, and one must know how to manage and operate users.

Therefore, products and operations are inseparable.

In fact, real user operations are divided into two categories:

One is a small-scale, relatively centralized operation. For example, when operating against dozens of KOLs, it relies on manpower, interaction, and emotions, and is generally used for the operation of key users.

The other is large-scale, strategic-based operation, which relies more on rules, mechanisms, and systems and is generally used for operations with large user bases.

This is also the part where product managers need to directly intervene to formulate corresponding strategies and even build systems. The more basic state focuses on the surface or single points, such as retention, activation, recall and other single links; while the advanced state focuses on the core and system building, such as user operation model sorting + user operation system building (including a large number of strategies).

There are many skills in small-scale user operations, which test the emotional intelligence and expression ability of operators. It requires a lot of practice to gain experience and insights. Every profession has its own expertise, so I won’t go into details here.

Large-scale user operations are strategic and can abstract standardized methods or processes and reuse them in different scenarios, which is more valuable for product managers.

1. What is large-scale strategic operation?

There are three principles for large-scale strategic operations: data-driven, refined operations, and automation (or productization).

1. Data-driven

We often mention data-driven in our work, but what exactly is data-driven?

First, we need to understand the three basic categories of data: business data, user basic data, and user behavior data.

Business data refers to data that is directly related to your business and has commercial value, such as recharge and investment data in financial products.

User basic data refers to the user's basic attributes, such as gender, age, region, occupation, etc.

User behavior data refers to the behavioral path that users take in your product when using it, such as function usage, page dwell time, etc.

These three types of data are interrelated. The key is to find the correlation between these three types of data, and then conduct behavioral analysis on specific user groups in combination with actual business scenarios. This is data-driven.

The problem here is data acquisition. Generally, data burying is adopted, which mainly has two types:

Either your company can inject code statistics into the product and build a corresponding backend for query;

Or you can consider connecting to a third-party data analysis platform, such as ZhugeIO, Sensors, Umeng, GrowingIO, etc.

The basic principles of these two tracking methods are the same: identify the user and then automatically collect information in the application.

Data is the basis of analysis. After obtaining the above-mentioned types of data, we begin to carry out refined operations.

1. Refined Operation

The so-called refinement is to segment users according to their differences and formulate different operation strategies for different user groups.

There are usually four dimensions to consider when segmenting users:

1. Population (gender, age, region, occupation)
2. Channels (natural traffic, channel delivery)
3. Context (time and place, such as commuting, before bed, weekends)
4. Usage process (access, registration, use of ABC functions, payment)

When segmenting users, pay attention to whether the user segmentation is reasonable: whether there are significant differences and regularities in user behavior.

After successfully segmenting users, it is necessary to formulate corresponding operation strategies. For example, for e-commerce products, if a novice user adds an item to the shopping cart but does not pay, a push or SMS reminder will be sent after one day. This is an automated strategy.

1. Automation (Productization)

When the scale of automated operation strategies continues to grow and become systematic, how should we manage them?

The key to controlling complex systems is to establish rules that support the operation of the entire system.

When the system is relatively simple, it can be broken down into multiple elements or links for management; when the system is relatively complex, it can be broken down or organized into multiple independent subsystems for management.

In our actual product work, we often use various operation systems, such as user life cycle management, user stratification, points system, task system, membership system, medal system, etc.

They can be summarized into three subsystems: user life cycle management system, user stratification operation system, and user behavior incentive system.

Subsystem 1: User lifecycle management system

We often talk about the user life cycle, but products also have a life cycle.

We often ignore the soil in which trees grow, so we must first clarify the product’s life cycle stage, understand the characteristics and development trends of this stage, and analyze the user life cycle based on the product life cycle.

• The product life cycle can be roughly divided into four stages: start-up, growth, maturity, and decline.
• The user life cycle is divided into five stages: introduction, growth, maturity, dormancy, and churn.

Combining the product life cycle with the user life cycle, different product stages have different user operation focuses:

In the start-up phase, the focus is on attracting new users; in the growth phase, attention should be paid to conversion; in the mature phase, loss prevention should be carried out; in the decline phase, new growth points should be considered and ways should be found to migrate users to new products.

After understanding the life cycle of your product, you can focus on operating it in combination with the user's life cycle.

1. How to build a user life cycle model

Building a user life cycle model is divided into three steps:

1. Sort out business logic (everything starts from the actual business scenario)
2. Find the key driving features that affect user retention or consumption (usually the core features of the product)
3. Define user behavior at each stage (combine data analysis to find key data indicators)

1. How to manage user lifecycle

Building a user life cycle is only the first step. The next step is to consider how to manage the user life cycle. There are two goals: to amplify the value of a single user and to keep users in the high-value range longer.

Amplify the value of individual users, that is, upgrade a user from A (such as the growth stage) to B (such as the mature stage).

We need to find the correlation and differences between different user groups A and B through data analysis. There are three analysis angles:

1. For users going from A to B, which path is the best?
2. Among users moving from A to B, what characteristics do most users meet?
3. Among the users going from A to B, do most of them have some similar behaviors?

After that, we will guide user A through some operational strategies.

Another scenario for using the user life cycle is to extend the effective life cycle of users and establish a mechanism to warn of user churn. This can be divided into four steps:

First, you need to define lost users. You can choose key behaviors (such as investment amount and frequency) or time length (such as the inflection point of the retention curve).

After defining the churned users, you need to analyze the behavior or attribute characteristics of this user group to find signs of churn, such as:

1. What similar behaviors did the user perform before churn?
2. Are users concentrated on a certain channel?
3. Basic attributes of users, such as gender, age, region, and occupation, are similar
4. At the time of churn, what actions the product took, such as whether a new version was released or whether key functions were changed

After finding clear characteristics, we can set up an early warning mechanism and complete user guidance.

Subsystem 2: User stratification operation system

In our actual work, we often mention user grouping and user stratification. There are differences between these two concepts.

User stratification is centered on user value. Under the same stratification model, a user will only be in one level; user grouping is centered on user attributes, and a user may have multiple attributes at the same time.

For example, users are divided into different levels according to their investment amount, which is user stratification; based on age and region, the same user can be in both the Beijing user group and the 20-30 year old user group.

The most commonly used user segmentation method is the RFM model: Recency (most recent transaction), Frequency (transaction frequency), and Monetary (transaction amount).

Based on these three indicators, the customer value can be described and divided into eight categories.

RFM is a way of thinking, not the only way to divide. You can find at least two key indicators for cross-analysis and finally achieve user stratification.

The RFM model can be used in different industries:

1. Finance: investment amount, investment frequency, and last investment date
2. E-commerce: purchase amount, purchase frequency, and last purchase time
3. Game: level, game time, recharge amount
4. Live broadcast: Watch live broadcast duration, reward amount, last watch time

Take community products as an example:

• Core content contributors (dark green): maintain deep emotional connections and share benefits;
• Secondary core content contributors (light blue): Try to transform into core content contributors;
• Mainstream content consumers (orange and yellow): do a good job of content push and ensure activeness; strengthen incentives and guide content production;
• New users/potential lost users (red): do a good job of content push to prevent loss;

Subsystem 3: User behavior incentive system

The highest level of managing user behavior is to form habits.

Points, memberships, medals, tasks, rankings, etc. are all incentives for user behavior. These incentives can be divided into three categories: short-term incentives (such as increasing investment amount within a certain period of time), single-point incentives (such as incentives in the card binding stage after registration), and long-term incentives (such as membership, levels).

Here are four common types of incentives:

1. Membership System

Members can be divided into three categories:

1. For new users, package a cost-effective product to improve conversion (operation method of traffic pool, no need to consider classification);
2. Target the highest value users to increase loyalty;
3. Make all users members (combined with a level points system);

To build a membership system, there are three steps:

1. Sorting out the rights and interests that can be given to users
2. Determine whether it is necessary to classify members (if the product volume is large or the user groups are very different, consider classification)
3. There are two main ways to rank players: rank players by spending money and rank players by growth value.

QQ membership has the longest history in the industry. It is a paid membership. Due to the huge number of QQ users, QQ members are also graded. Members of different levels have different rights and interests, and there is a clear distinction between them.

1. Competitive ranking

WeChat Sports is a great example of competitive rankings, which fully utilizes its social attributes, which can have a greater motivating effect on users compared to people they know in their social circles.

WeChat uses the simple behavior of daily step count as a reference, and pushes it out on a daily basis. Users can get timely feedback and see their own step count and ranking. In the leaderboard, the user ranked first can occupy the cover of his friend, which is also a very attractive honor incentive.

1. Identity Honor

Identity labels or medal systems are generally used in community products to stimulate users' sense of identity honor and increase user activity. Nodes are generally located on the user's necessary behavior path.

Take Zhihu's medal system as an example. A series of medals are set up on the user's necessary path: complete personal information, follow 20 people, comment 10 times, agree 100 times, log in for 100 days in total, ask 5 questions, answer first, answer 20 questions, be followed by 500 people, get 1,000 likes, become an excellent answerer, columnist...

We can see from this series of medal settings that Zhihu is gradually cultivating users' behavioral habits, guiding a novice user from a content consumer to a content contributor and then to a core content contributor.

1. Points system

This is a relatively complex incentive method. There are three common reasons why many products fail to achieve points:

1. No stable budget source
2. Imbalance in points earning ratio
3. The consumption of points is unstable, and there is even no channel for consumption of points.

First of all, we need to consider the budget issue. There is no way to earn points without a budget. This is the basis for earning points, so if the boss doesn’t provide a budget, you can just forget about it.

When considering the points budget, you have to consider it as a whole. The product has a total user retention system budget: in addition to the points budget, there is also a budget for membership benefits, a budget for preferential subsidies, a budget for other benefits, etc., so you have to first look at the proportion of the points budget in the overall budget.

When setting a points budget, there are two common types:

1. Fixed amount method: often used for non-direct payment products; for example, setting the overall budget for points to 1 million
2. Fixed ratio method: often used for direct payment products; for example, setting a 1% commission on the user's investment amount as a budget

After solving the budget problem, we need to consider the overall construction of the points system.

The points system actually has economic principles in it. What is often discussed in economics is the relationship between supply and demand, and the same is true for the points system. It can be divided into points acquisition (supply) and points consumption (demand).

Points acquisition (supply)

If points are too easy to earn, it will cause inflation and currency depreciation, which means the depreciation of points. Because costs need to be controlled, more points are needed to redeem goods while keeping the value (cost) of the goods being redeemed constant; if points are too difficult to obtain, it will dampen users' enthusiasm and fail to serve as an incentive.

Points consumption (required)

If points are consumed too quickly, it means that the redemption ratio of points to goods is too low or the redemption threshold is too low, making costs difficult to control; if points are consumed too slowly, it means that the redemption ratio is too high or the redemption threshold is too high, which will also affect users' enthusiasm for obtaining points.

After going online, if the redemption ratio or redemption threshold is adjusted, it is easy to cause user dissatisfaction and lead to a large number of customer complaints.

Therefore, when building the points framework in the early stage, it is necessary to do a good job of calculating the points cost and control the cost of individual users and the overall cost.

The design of points acquisition is divided into five steps:

(1) Behavior analysis

Based on the actual business process, user behaviors can be divided into core behaviors and non-core behaviors. For example, in financial products, recharge and investment behaviors are core behaviors, while check-in, information browsing and other behaviors are non-core behaviors.

(2) User stratification and points ratio

Users are stratified according to the quantitative values ​​of their core behaviors, and the points earning ratio for core behaviors and non-core behaviors at different levels is set. It is important to reserve points for future online launches that require guiding user behaviors and points issued through activities.

(3) Calculation of total points for a single user

• Before calculation: obtain points according to different levels of users and calculate according to the user's maximum points.
• After calculation: adjust the ratio based on the set integral ratio of core behaviors to non-core behaviors.

(4) Calculation of total user points

Calculate the maximum points that can be obtained and the predicted points for users at different levels. When calculating the predicted points, set a preset acquisition rate (based on historical data) for each behavior.

(5) Correction of points redemption ratio

• Calculate the points redemption ratio: Points redemption ratio = total user predicted points/total budget
• Points redemption ratio adjustment: Redemption ratio conversion factor = target redemption ratio / current redemption ratio
• Adjust behavior scores: Single behavior scores are non-integers and can be adjusted within a small range
• Calculate the cost per user: Calculate whether the subsidy cost per user at different levels is reasonable

Regarding points consumption, planning can be done from the perspective of channels and content: channels include points malls and points activities (lotteries, games, etc.); contents include closed-loop products (coupons, red envelopes) and open-loop products (purchasing goods, self-production).

After completing the planning of points acquisition and points consumption, you should also pay attention to the maintenance of points in actual operations.

The health of the system is tested based on data indicators. Common indicators include: predicted amount of points to be issued, actual amount of points to be issued, point consumption rate (points consumption/actual amount of points issued), and percentage of people who consume points (number of people who consume points/number of people who currently have points).

• If the actual amount of points issued is greater than the predicted amount of points issued, it means that it is relatively easy to obtain points and the threshold for obtaining points should be raised;
• If the actual amount of points issued is less than the predicted amount of points issued, it means that it is too difficult to obtain points. The threshold for obtaining points should be lowered or more channels for obtaining points should be added.

It is generally reasonable for the points consumption rate or the ratio of points consumption to the number of people to be above 40%. Of course, the higher the better. If it is too low, you should consider adding consumption channels or adjusting the content of the redemption products.

Summarize

The above introduces the three subsystems in the user operation system. You may have a systematic understanding of the various operation systems you encounter in daily life, but here is another question - regarding these three subsystems, user life cycle management system, user stratification operation system, and user behavior incentive system, when should which incentive method be used?

In fact, it can be analyzed in combination with the product life cycle:

A. During the introduction and growth stages of the product, you can do simple user lifecycle management and provide periodic user behavior incentives, because at this stage the main consideration is to attract new users, that is, incremental users.

B. In the later stage of the product growth period, when the number of users begins to slow down and the total number of users of the product gradually stabilizes, it is necessary to start building an overall user stratification system and carry out refined management.

C. During the maturity and decline stages of a product, the focus should be on preventing user churn, which requires more sophisticated user lifecycle management and a site-wide user behavior incentive system.

If the product you are making already has these different operating systems, and they may be managed by different departments or different people, then you need to consider how to integrate these operating systems and carry out collaborative planning.

At this time, it is necessary to re-examine the product's actual business processes and user growth paths, refer to the user value stratification method, and build a new user growth ladder and operation model.

To put it simply, it means sorting out, integrating, and rebuilding user levels, then putting them online for operation, and constantly adjusting based on actual results.

Related reading:

1. APP promotion and operation: How to maximize the effect of your activities?

2. APP promotion planning: 60,000 paying users increased within 7 days of beta testing!

3. APP promotion activities: How to plan a screen-sweeping event?

4. A complete list of APP promotion methods in 2019, take it and don’t thank me!

5. How to carry out APP promotion and marketing? What are the common methods?

6. APP promotion case: How to go from 0 to millions of users?

Author: Yue Xiaoyu

Source: Yue Xiaoyu