The basic skills of user operation: activation and retention

Why do I call it basic? We know that the AA AR R model of operation corresponds to the user's life cycle. In the structure of most Internet companies, acquiring users is the job of the promotion and channel departments, and the next actions are what our operations must perform. However, user conversion to payment and user self-propagation are later behaviors. Therefore, I consider user activity and user retention as the basic skills of user operations .

User Modeling

User modeling is a necessary preparation before launching many user operation methods. User modeling, in simple terms, is to classify users into various categories based on their attribute characteristics. Generally speaking, similar people have similar habits and behaviors. When operating a large number of users, user modeling is the basis for refined operations.
User Attribute Model

User attribute modeling is to classify users based on their own characteristics. Generally speaking, the classification methods include user's age, gender, social class, family composition, personality, and lifestyle.

Here is a brief explanation of the following classification concepts. Social classes are mainly divided into top social figures, social elites, middle class, well-off class, working class, lower class and people in difficulty based on people's educational background and disposable income. Family composition is easy to understand, including three generations living together, young couples, single-child families, two-child families, etc. Personality can usually be classified into self-confidence, control, autonomy, submissiveness, defensiveness, adaptability, etc. The classification of lifestyles is more diverse, such as Lohas, environmentalists, etc.

In short, user attributes refer to the personal characteristics of the user. They are some essential attributes of the user that are not easily changed by changes in product form. In traditional marketing methods, we regard this behavior as consumer analysis.
User behavior modeling

User behavior modeling is to model users based on their behavior when using our products. Unlike user attribute modeling, which focuses on the users themselves, user behavior modeling only generates user classification based on the interaction between users and products.

This modeling method has many more dimensions. For example, the channel from which the user is introduced, whether the user has uploaded an avatar for himself, the average order value range of each purchase by the user, the daily usage time of the user, etc. All user interaction data can be used as dimensions for user modeling.
Defining user retention and user activity standards

Now that we understand the basic ideas of user modeling, we can start to activate and retain users. Retention and activity are both imprecise adjectives. The definitions of churn and activity will be different for different products. For example, we may open Onmyoji and play a game every day, but it is difficult to stick to visiting JD.com and placing an order every day.

“Churn behavior is a long-term and continuous behavior (retention is also the same. It does not mean how long the user has used the app, but that the user has been using the app continuously for a relatively long period of time).”

Therefore, the definition of retained users can be in the following format: users who have not logged in (browsed, ordered, paid, depending on the platform) for X consecutive days (weeks, months) are called lost users. Similarly, the definition format of an active user can be expressed as a user who has logged in (browsed, placed an order, paid, depending on the platform) for Y consecutive days (weeks, months). And X>Y, because I think there are always some users in the middle state, that is, not too active, but not to the point of churn.
How to find this X, Y, and define user churn and activity

As mentioned above, user modeling is a necessary preparation before implementing many user operation methods. To accurately define user churn and activity, it is necessary to use a user model. Here we need to model the user's behavior and build a model based on how many consecutive weeks a user has not repurchased.

Let’s take a look at the example directly. The following chart simulates the user purchasing trend of an e-commerce product. If 1,000 users place an order on the platform on the first day, there will be a user model of how many users will not repurchase each week.

We can see that after the third week, the number of users who did not repurchase has not decreased any further. If no measures are taken, this group of users will continue to have no repurchase trend until the tenth week. As shown in the figure above, we can define that if a user does not come back to repurchase within 3 weeks, he or she can be considered a lost user, that is, X in the above definition is three weeks.
User retention: Find users who are about to churn

In the above text, we have found that if we do not take any measures, the user churn rate after the fourth week will be as high as 40%. However, we only know that it is 40%, but we don’t know which 40% of these 1,000 users will eventually churn after we get another 1,000 users. Here, we need the idea of ​​user modeling again. For example, after our analysis, we found that half of these 400 users came in through a certain channel; or they are all in the same age group; or, this group of users has never bound their Alipay accounts after registration. We are always able to draw a portrait of this group of users who are prone to churn through multi-dimensional user modeling.
Loss warning mechanism

Based on the user behavior model, we found some common points of lost users. What behaviors did these users do before they lost? For example, we found that users will churn after using a large-denomination coupon, or after not returning for N days. For users with similar behaviors, we need to communicate with them about churn warnings to prevent user churn, because once users churn, it will be extremely difficult to recall them.
Communication to prevent loss

When we find a group of users that meet many churn warning models, we will take some anti-churn operational methods to prevent user churn.

The core idea of ​​preventing churn is to meet user needs, find new needs that users may have, and tell them that you can meet their needs. The first two often depend mainly on the product form, and what operators mainly need to do is to "inform users that your product can meet their needs." There are many channels to reach users. SMS, EDM, APP push and social media are all communication channels that can reach users.

To ensure effectiveness as much as possible, we recommend using multiple channels to reach users when resources are sufficient. However, to avoid excessive harassment of users, we need to pay attention to channel deduplication when sending communications to the same user. In my experience, we usually use social media push first (such as WeChat text and picture messages and customer service information), and then use APP push for those who cannot be reached. If the users cannot be reached, we will use SMS push. Since EDM is PC content, it can be operated independently.
Content Strategy

Regarding the wake-up copy for churn warning users, there are mainly the following points to note:

（1）Consider the user’s reading context

Generally speaking, when a text message is received, users will spend less than a second to scan it. Our goal is to allow users to get the main information we give them in this glance. Therefore, considering the user's reading scenario, when writing APP Push and SMS, we should be careful not to use uncommon words, try to use short sentences, and use the [ ] symbol to indicate discounts and other preferential information. When pushing APP PUSH, you can also add emoji expressions to attract user attention.

In the picture above, what catches the user's eye are long sentences and complex rules. The user may close the text message before getting your message.

(2) Be sure to tell users the benefits of returning to use or placing an order immediately

Interest points are baits that drive people to take actions that go against their own thoughts. I won’t go into details here. This is why, in the era when Didi was burning money like crazy, you would receive Didi coupons if you stopped taking a taxi after a period of time. If you are using coupons as bait to lure users back, remember to tell them.

Excellent cases have clear benefits + short sentences + sense of urgency + proper nouns enclosed in <>.

(3) Recall users based on the reasons they came to our platform

I don’t know when I posted a job advertisement for someone else’s rice noodle roll shop. After logging into 58 for a while and becoming a lost user of the platform, I received a text message like this. For social platforms, you can tell users how many of your friends have updated their status during the time you were away. For recruitment platforms, you can tell users that there are people submitting resumes for your positions, so that users will want to come back to your platform to take a look.
Incentives to activate users

Just like winning back users who are about to churn, increasing user activity also requires using interest points to motivate users. Let me share with you an incentive for user activation that I experienced when I was working on an e-commerce platform.

When we were conducting user modeling, we found that if a user purchased a second order within 7 days after receiving the first order on the platform, the probability of the user becoming an active user of our platform in the next six months would increase by 120%. Therefore, we believe that this user model of placing another order 7 days after receiving the first order can effectively improve user activity. Therefore, we decided to allow users to enjoy free shipping within 7 days after receiving the first order.

This is a means of user motivation, which incentivizes users through some interest points so that ordinary users can try to get closer to the user model of active users.
Systematic user incentives

In many products, we can see point systems , medal systems, achievement systems, etc. The core logic of these systems is to give users tangible points or intangible medals or achievements as incentives after completing one or a series of actions.

Regarding building a user incentive system, I would like to briefly talk to you about Professor Jin Pu’s four steps of user incentives.

To find the object of incentive, such as a content platform, it will involve content producers and content consumers, and the incentives for them are completely different.
It is necessary to clarify which behaviors need to be incentivized and which behaviors should be prohibited. For example, on Weibo, frequently logging in, posting on Weibo, and following each other are all behaviors that need to be encouraged, but spreading rumors and slander should be prohibited.
It is important to clarify what kind of benefits will be used to motivate users. In addition to conventional material accumulation such as points and cashback, we can also use some incentives that satisfy the human nature of users. For example, medals are for showing off and vanity, and privileges (such as senior Baidu Tieba members can sign in to multiple bars at a time) are for satisfying users' arrogance and vanity.
Choosing a reasonable way to formulate reasonable rules means choosing the appropriate incentive method based on the user behavior ratio engineering required for a healthy product form. We call it adapting to local conditions.

The user incentive system is a huge systematic project, and operations need to work together with products and development to build it. This time, the author will give you a basic introduction, and will discuss it in depth with you through other articles later.
Final Thoughts

The routines for user activation and retention are as mentioned above. You may understand the routines at a glance, but it still takes a period of work experience to actually use them, especially how to model users, how to write copy to recall users, and how to build a user incentive system, which cannot be mastered overnight.

The above are some summaries of my work experience and essays written after exchanges of ideas with some giants. If there is anything inappropriate, please point it out. I hope that the great masters will help me continue to grow.

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

The author of this article @Haohao Operation Compiled and published by (Qinggua Media), please indicate the author information and source when reprinting! Site Map