Detailed explanation of the three major issues of user activity, retention, and loss

Before solving a problem, do you want to understand what the problem is? It is said that user retention , user activity, and user churn are three difficult issues to deal with. In fact, we may not even know how these three are reflected in the data.

Many students complain that the three major issues of user retention, user activity, and user churn are difficult to deal with.

First of all, these three questions seem to be the same, but not exactly the same, so it’s not quite clear what they are.

Secondly, these three issues are often interrelated and influence each other, and I feel dizzy while talking about them.

Thirdly, it is difficult to give analytical suggestions on these three questions. Once the numbers are calculated, we often don’t know what to say.

Every time I do an analysis, I can only say: "Aim high! Aim low!" or I write down timidly: I have to communicate with the business/ask the users for the specific situation. I can't give any constructive suggestions. Today we will first clarify the origin and figure out what these three are.

1. What do the three mean?

User retention, user activity, and user churn are actually three different statistical methods of user active behavior.

1. User activity

First, we define “active” behavior, such as logging in once, visiting for more than 10 minutes, making a purchase once, etc. After that, each time the user performs an active behavior, it is recorded as: active once (as shown in the figure below).

2. User Retention

If the user continues to be active after a period of time starting from the specified time, it will be counted as one retention. The most common is new user retention (as shown below).

3. User churn

Artificially define a time point as a churn node, such as when a user has not logged in for 12 months. Those who reach the node are considered lost users.

Note that unlike user activity and user retention, user churn is not an objective fact but a subjective determination. Theoretically, as long as the company does not actively cancel accounts, you can assume that users will never be lost. Of course we know that this is self-deception, so we usually give a specific loss standard.

2. The connection and difference between the three

Students with quick reactions have already noticed that there is an obvious connection between these three.

The relationship between user activity and user retention is as follows:

The relationship between user retention and user churn is shown in the following figure:

So, if the activity criterion is logging in, and the churn criterion is not logging in for three months:

1. Xiaoming registered on February 1 and was counted as an active user of that month
2. Xiao Ming did not log in in March, April, and May, and was counted as an inactive user and a non-retained user.
3. Xiao Ming still hasn't logged in in June and is counted as an inactive user, non-retained user, or lost user.

Since all three indicators point to active behavior, why should they be divided into three?

Because these three indicators actually represent three directions in which the business can act:

1) User activity is an indicator that can be counted in real time.

Therefore, actions in the short term will immediately reflect on it. Positive actions, such as big sales and promotions, can immediately see feedback; negative actions, such as server downtime, can immediately see the impact.

2) User retention requires a longer period of observation.

Therefore, it can better reflect systemic and structural problems. For example, poor product experience, insufficient competitiveness, inadequate operations, etc.

3) User churn is the bottom line for saving users .

The longer a user has not come, the higher the cost of recalling them, and the user may even have forgotten that there is such a product. Setting churn indicators can better remind you: how many users have reached the bottom line, pay attention!

When doing business, it is important to use a combination of long and short sticks. The three indicators of user activity, user retention, and user churn point to the short-term, medium-term, and long-term initiatives of the business. Therefore, separate statistics and analysis are needed. In fact, when the activity of the user group decreases, these three indicators often move together, making analysis difficult. However, it is not just numerical calculation issues that make the analysis difficult. The following are the real troubles.

Why is analysis difficult?

The reason why it is difficult is this: if the user logs in, we know exactly what he did after logging in, but if the user does not log in, we do not know why he did not log in . The reason for not logging in can only be guessed. This is the core of the difficulty of these three problems. If the leader asks us, "Which pages did the logged-in users visit today?" I guess everyone can answer fluently.

But if the leader asks:

• What is the reason for inactivity?
• What is the reason for the low retention rate?
• What is the reason for the increase in churn rate?

I guess many people have black lines on their faces. If you add another sentence at the end: "Tell me the specific reasons! Don't just talk about numbers!" I guess many people would collapse. All I can think about is: "I'll ask the product manager," "I'll talk to a few users," and my hands are probably already reaching for the phone...

However, this is only the first level of trouble; the second level of trouble is even more difficult to deal with. Apart from super apps like WeChat, there are no other apps that users must open every day.

• Some applications have a very obvious window period, such as renting a house and traveling;
• Some applications are just naturally inactive, such as Internet financial products;
• There are some applications that you love to death, and those you don’t love will be thrown away immediately, such as games, short videos, and social products.

So a concept is derived: the natural life cycle.

• The natural state of users is to log in during winter and summer vacations
• The natural state of users is to churn after using for XX days.
• The natural state of users is to log in once or twice every quarter

The difficulty lies in the word "natural" .

You will find that this "natural life cycle" is like air. Everyone knows that it exists, but they can't see or touch it. Who says there are only so many natural cycles? What reasons do you have to be sure that these users are naturally lost rather than the product is not doing well? It seems that this value exists, but when it comes to a specific user, it seems that it does not exist.

So, just like the natural growth rate, the so-called natural life cycle can easily become the main battlefield for operators to pass the buck.

The high activity rate and retention rate are my achievement. The low activity rate and retention rate are natural reasons. If I don’t do anything, it will be even lower. This kind of argument will directly lead to the boss's dissatisfaction and also lead to higher requirements for data analysis - everyone hopes that the data analyst can clearly say: Zhang Zihan left due to natural reasons, he was very satisfied with our product and just stopped using it naturally, there is no other reason. Hmm, do you really want to check Zhang Zihan's data cable to read his brain waves directly?

However, this is only the second trouble; the third trouble is even more difficult to deal with. It's just that everyone instinctively thinks that as long as they figure out the reasons for user inactivity, low retention, and churn, they can win users back. This is a very futile and stupid idea, just like when breaking up, the boy grabs the girl's hand and asks bitterly, "Why are you leaving me, why, why, why is this, give me a reason"

Anyone who has watched two episodes of a romance drama knows that this is meaningless. But when it comes to myself, I always can't help but ask, "Why, why, why is this!" The only difference is that this time it is the boss who asks. You can't just go to a late-night snack stand, order a few beers and a bunch of skewers, and get him drunk like in romance dramas.

The above three problems make it easy to calculate the three values ​​of user activity rate, user retention rate, and user churn rate, but it is very difficult to analyze the reasons behind them and find targeted countermeasures. Especially when the user activity rate decreases, the user retention rate decreases, and the user churn rate increases, the performance pressure will cause the motivation of various departments to blame each other and make unreasonable demands to increase sharply .

If students who work with data want to get out of this situation safely, they must keep in mind the principle that "truth is the first to die on the battlefield", carefully analyze the pros and cons of specific problems, and find solutions.

Author: Down-to-earth Teacher Chen

Source: Down-to-earth Academy