How to analyze the loss of users?

A student asked: User churn How to analyze? The data of user churn rate can be calculated. What will happen after that? It seems that there is no reason for the loss just by looking at the data. It only knows that the user has X I haven't come for months, and I don't know what I can do when I see this. Today, the system will answer the question.

one

Common errors in user churn analysis

▌ error one ： Try to retain every user. This is operate The most common mistake, many newcomers will step into this hole. When you stop shopping, you will be issued coupons, and you will not log in to the carousel. As a result, they burned the funds empty and raised a group of unprofitable wool customers who could not get up early. In fact, the loss of users is inevitable 100% Retention of. Every business should pay attention to its core users. When it comes to user churn, what we really need to do is to keep the churn rate in a cage and control it at an acceptable level.

▌ error two : Try to understand every reason for loss. This is the most common mistake in analysis, and many newcomers will step into this hole. Users don't like it? We didn't do it well? The opponent is too powerful? The user has no money—— In short, I want to give everyone a reason. But there is no data at all, so big eyes stare at small eyes. In fact, we do not need and cannot enumerate all the reasons. As mentioned above, we only need to control controllable factors and reduce obvious errors.

▌ error three : Focusing only on the loss rather than on the activity, we should be wise after the event. This is another common mistake. The analysis was started after the loss rate actually increased. As a result, the user has run away and the analysis is useless. The loss rate is a lagging indicator. Before the "loss" of data, users may have run away, and have not been active for several months. Therefore, the turnover rate should be combined with the activity rate. Pay attention to events that affect user activity as soon as possible, and closely track the activity rate of core users to avoid doing useless work afterwards.

two

Basic thinking of user churn analysis

The goal of user churn analysis is to keep the churn rate in a cage, so in terms of data, we first focus on the trend of churn rate, especially on three types of problems (as shown in the figure below).

one Event type problems. By once / Short term loss rate fluctuation caused by multiple events.

two Systematic problems. The overall turnover rate of the company is higher than that of peers / The level of experience is high.

three Continuous problems. The loss rate has continued to increase from a certain time, and there is no sign of improvement.

Loss rate is a concept relative to activity rate. Although we are used to giving a "user X Don't log in every month / No purchase ” This is the definition of lost users. However, when users are inactive, the actual loss may have occurred.

In order to better find the loss problem, we often look at the natural cycle and life cycle in combination with the activity rate. The natural cycle often points to the event type problem (because the event occurs on a natural date), and the life cycle often points to the system type problem (the business is not well done, and the user's life cycle is short or There is a breakpoint).

three

Event based problem analysis method

Negative events can lead to user churn. Such as out of stock, price increase, system BUG , user complaints, competitor promotions (we just haven't done that yet), etc. This type of event is most easily identified. Reflected in the data, the activity rate of the user group affected by the event will fall after the event N The turnover rate began to increase in the past six months.

In this way, it is easier to see the results. When designing retention methods, it is easier to suit the remedy to the case. Finding the real reason that makes users unhappy is better than simply inserting coupons to retain users.

Note: Positive events will also increase the turnover rate. In particular, users pull new, promote activation, retain, awaken, and so on. Simply stimulating non consumer soft indicators is most likely to lead to false prosperity.

Objectively, as long as there are preferential activities, they will attract arbitrage customers, and such users have a high natural turnover rate.

Subjectively speaking, in order to create good looking data, operators will also reduce restrictions and leave arbitrage space. The effect of positive activities is often discounted by two actions. For example, for new user registration, the user life cycle churn rate generated by the pulling of new users is easily higher than that of normal new users (as shown in the figure below) N This batch of users will inevitably have a high churn rate in six months.

Therefore, when doing activities, we must consider the relevant consequences in advance. A positive event is different from a negative one. There is still work to be done. It would be better if we comprehensively evaluate it. Although the final result may be something that the planning and operation do not want to face, what is actually investigated here is everyone's integrity.

four

Systematic problem analysis method

If there is a systemic problem, just say one thing: our business is worse than that of our competitors. At this point, diagnosing business problems and improving business performance is the core. The diagnosis method can refer to the user life cycle theory.

The reasons for the loss of users in the entry period, growth period and maturity period are different, and the focus of analysis is also different. In order to save space, here is a brief summary as follows. Interested students can click at the bottom right corner at the end of the article to see more than 60 Let's share it again.

When dealing with systemic problems, different stages have different priorities.

▌ Entry period: Generally, in the entry period, there will be no difference in improvement. In the entry period, users have not actually experienced the core selling points we provide, so we need to improve the process indiscriminately, so that users can experience the core selling points as much as possible.

In the Internet industry, people tend to focus on the process of novice tutorials in Black One Minute (one minute from download to registration). In traditional industries, it is often emphasized to use greeting scripts to let users have an experience and try the product as soon as possible.

▌ Growth period: After entering the growth period, it needs to be treated in categories. After entering the growth period, marginal users and wool users will be eliminated, and the user value will start to differentiate. Non core users should be lost. It is just a waste of money to retain them. It will also devalue the brand because of frequent discounts.

At this time, we should pay special attention to the loss of core users, the decline in the activity rate of core users, the shortening of their life cycle, and the decline in the proportion of core users among new users. These are major problems that need to be carefully sorted out and solved. It is possible that we have already started to take action before the drain really rises.

Systematic problems may not be solved in one step, but in a continuous iterative process. It is possible that we can diagnose the problem, but the solution is not easy to use and cannot improve the data. Therefore, if systemic problems are found, you need to:

Finally, what we see is that our user retention curve is getting closer and closer to our competitors, and the churn rate continues to decline. At this time, we can say that the systematic churn problem has been solved. There may be many tests and attempts in the process, so it is necessary to make observations and records to fight a protracted war.

five

Continuous problem analysis method

Persistent problems are often the most difficult to solve. In fact, the loss rate, activity rate, retention rate and other data often fluctuate irregularly and slightly, rather than continue to grow significantly.

This is the real chicken rib problem: Leave it alone and the leaders always ask. If you want to manage it, you have no clue. Even the churn rate has risen for several days, and it fell back before the analysis report was written. It's really embarrassing.

The processing sequence is event type, system type, and continuous type. Because a single major event is most easily identified and can be easily seen through data. At the same time, a series of events are often the root cause of systematic and continuous problems, and it is also helpful to identify specific events to deal with other problems. Systematic problems are relatively easy to deal with because they can find appropriate benchmarks when the business side is experienced.

The most difficult problem is the persistent problem. The change in the churn rate usually does not last to be particularly serious, but fluctuates repeatedly in a small range (as shown in the figure below). In the absence of experience and data accumulation, it is difficult to fully identify these small fluctuations, so the final solution is to solve them.

If it can't be solved, set up observation indicators and track them first. To a certain extent, we may find clues.

six

Differences in loss processing of different business types

Because the churn problem is highly related to the business, different business churn analysis directions are also different. In terms of the general category, there are two most important distinguishing dimensions.

▌ Valuable low frequency product VS Cheap FMCG.

The more expensive the product (car, house, big home, wedding...), the longer the user's decision-making process, the more inclined to make judgment in advance, and there is no repurchase theory. There is an obvious window period for such business users to make decisions deadline The closer it is, the more likely it is for the user to make a final judgment.

Therefore, the loss of such business users is a countdown hourglass. At the first time of contact with users, it is necessary to understand the user status: what the user needs, which competing products have been compared, and whether the price has been negotiated.

In this way, we can roughly judge how much time we have left. So as to better seize the opportunity to close a deal and follow up quickly. Instead of being silly and indiscriminate, we introduced and followed up step by step, and the day lily became cold.

FMCG, or consumer products (such as clothes, shoes, and mobile phones) with high purchase frequency are inherently low in loyalty, and are easily changed by popular trends and promotions. We can adopt the strategy of no gap retention. Anyway, the user won't buy it this time, and will come back later.

Therefore, when dealing with such products, Internet enterprises often distinguish between platform loss and product loss.

As long as users stay on the platform, they will continue to wake up. Traditional enterprises often use seasonal changes, new product launches, weekly celebrations, festival activities and other means to activate users more frequently. In short, as long as the user value is large enough, we will not abandon or give up.

▌ Traditional industries VS Internet industry.

The amount of data accumulated in the user life cycle of the two is different. There is a lot of data in the Internet industry, which can often record users from clicking the promotion link – Landing page – register – browse – The whole process of placing an order.

Therefore, funnel analysis is often used to see what steps lost users will get stuck in, and to lock the problem points for improvement. Especially in the registration stage of newcomers, the optimization is often undifferentiated.

Traditional industries often only have consumption data, so users can only be measured by consumption frequency and consumption interval. Ordinary users are consuming n The next time, the ones you don't like will be lost, and the ones you like will continue to buy. This is called magic number.

By comparing the size of magic numbers, you can know the gap between yourself and your opponent. As for the user's arrival at the store – welcoming – experience – service – There is no data at the behavioral level such as evaluation, which needs to be supplemented through market research and other means.

This is mainly a reminder that there are great differences between businesses, although the definition of churn can be defined as XX Don't log in every month / Do not buy. However, the actual loss scenario may have occurred long ago, and the key actions to stop the loss may not have data records. Think more about the method based on specific business, which is more effective than mechanical code.

seven

Summary

Many students find it difficult to deal with the problem of user churn. On the bright side, it is because there is little data lost by users, and we don't know what users think.

Essentially, it is because of the reasons that will lead to the loss of users, such as user life cycle, user clustering, user decision-making process, user growth path, new user transformation process User experience , User MOT , competitive products and other factors.

Any topic here can be put in a separate article. Having understood these, we basically understood the whole user operation process. In essence, it is difficult to analyze user churn. The difficulty lies in that the students doing the analysis seldom understand the business operated by users.

The answer is: I don't know. Even: I don't know anything. You ask him what he knows? He only knows how to calculate the turnover rate, and then makes a lot of cross tables according to the user's age, gender, registration channel, purchase frequency, etc. Then, we can face a set of data 1% 、 2% 、 3% What does it mean?

The above is a joke. In short, the analysis is not only to run a data table, but also to go deep into the problem and find the real root cause of the business. This article has been very long, and there are some points that are not detailed. I will fill them in later.

The original text originates from WeChat official account (Earthtouch School)