This is how products with over 100 million users retain users

The content of this article is a summary after actual trading and is highly reusable.

Anyone who works in growth must have heard of the “reservoir theory,” which says that the number of users of a company is like a large reservoir, with a pipe on top that continuously fills it with water, and a pipe below that lets the water flow out. The reservoir will never be full, which sounds like a waste.

The upper pipe here is called "attracting new users" and the lower pipe is called "churn". Many companies only know how to attract new users when increasing users, spending manpower and material resources like running water, but the water in the reservoir is getting less and less. This is because they do not understand the principle of increasing revenue and reducing expenditure. If you want the reservoir to be full, you must first block the lower pipe, the tighter the better.

Therefore, before you start attracting new customers, you must first think about how to retain them.

01 Before doing retention, you must first clarify what the value of retention is?

1. Make every penny worth spending

A product I made two years ago costs more than two million yuan per month to promote. For every 1 percentage point increase in product retention, I can save tens of thousands of yuan. And this 1 percentage point increase may be just because of optimizing a sentence of copy, so please remember that money is in the details.

Think about those products that spend tens of millions or even hundreds of millions of dollars on channel fees every month. Every one percentage point increase in retention will contribute huge value.

2. The compound interest brought by retained users

First of all, retention should run through the entire user life cycle. Every touchpoint that the user comes into contact with the product (product process, pop-up ads, guiding copy...) will determine whether he will stay, whether he will trust the product, and whether he is willing to recommend this product to people around him.

If you successfully retain the user and impress him, he will be likely to continue recommending the product to people around him, and your user pool will have a steady stream of free water flowing into it.

02 So how should retention be achieved?

As mentioned earlier, user retention cannot be limited to a certain link, but should be systematic and run throughout the entire user life cycle.

Let’s take a look at the AARRR model: Acquisition, Activation, Retention, Revenue, and Referral. Each link has countless points that can be optimized. Let’s sort out how to systematically retain users.

1. Path design

The design of user paths can be divided into three steps: setting expectations, focusing on reality, and adjusting the gap.

1) Set expectations

It is particularly important to do a good job of expectation management when doing retention. You must start from the end, think clearly about your purpose first, and then design the user path to ensure that new users use the product according to the established steps after entering the product. The purpose of doing this is to allow users to find the value of the product as soon as possible.

Users will stay because the product is valuable to them.

Ask yourself this question repeatedly: What information do users want to see in what scenario? What problem are you here to solve?

If new users do not receive clear guidance after entering the product, they may find that the product does not solve the expected problem after operating it for a few times, and then directly exit and uninstall it.

If the product logic is particularly complex, you must design multiple paths to accommodate users with different needs. The more thoughtful the consideration, the better the retention effect.

2) Focus on reality

Ideals are full, but reality is skinny. After designing the user path, everything is not over yet, because in practice you will find that users will use the product through various paths and will even lose users in various strange ways.

Maybe the product is too difficult to use, maybe it didn’t meet expectations, or maybe you were attracted by other information and jumped away and couldn’t find it again.

Taking e-commerce as an example, from starting, to selecting products, to adding to the shopping cart, to filling in the delivery address, to payment, this main chain of users will be diverted by various pages and promotional activities.

Traffic is like a river. As it flows forward, it will be divided into multiple tributaries. In the end, only a few of them can reach the destination. We must present all the tributaries in the form of data and find out the nodes that cause the loss.

3) Adjust the difference

After finding the nodes of loss, it is necessary to modify the originally designed user path.

Taking mobile phone number login as an example, the traditional login method is as follows:

1. Step 1 Enter your phone number
2. step2 Click to send SMS verification code
3. step3 Enter the verification code
4. step4 click the login button

A slightly better optimization would be:

1. Step 1 Enter your mobile phone number and the system will automatically trigger the verification code after verification.
2. Step 2 Enter the verification code. No need to manually trigger the login button. The system will automatically trigger the login after verification.

The optimization is as follows:

step1 Automatically obtain your mobile phone number, click to log in, done!

A login action can be simplified from 4 steps to 1 step, and what is reduced in the process is the traffic loss.

If you don’t know how to find the difference, remember a key word called “touchpoint”. Touchpoint means that we break down every page and every link in the user’s product usage process into pictures, texts, buttons, videos, animations… these are the smallest units of elements. Each touchpoint may produce a difference.

The picture below is a recently optimized WeChat sharing card. The front one is the original plan, and the back one is after optimization. It is obvious which one has a better effect.

2. User stratification

The purpose of user segmentation is to operate the traffic within the site more accurately and avoid the situation of grabbing everything at once. User segmentation can be started from two dimensions:

1) User’s personal attributes

It mainly includes gender, age, region, income, occupation, and device model. These six points constitute the basic layer of user stratification.

2) User behavior attributes

It mainly includes user activity, length of stay, depth and breadth of product usage, and the number of times different behaviors are triggered. These four points constitute the dynamic layer of user stratification.

In actual operation, the dynamic layer is more important.

Taking e-commerce as an example, the three most commonly concerned indicators are consumption frequency, consumption amount, and time of the most recent consumption. These three indicators can be used to stratify user value and activity. These stratification results determine the output of our final operating strategy. For example, the activation strategy, the L30-, L30+, and L60+ marketing and subsidy strategies must be incremental, and each level can be divided into n small levels.

Make a good framework, break it down into several layers according to the corresponding indicators, and make the granularity as fine as possible.

3. Refined operations

Before taking any operational action, you must first think clearly about the purpose, is it to increase DAU? Is it to increase GMV? Or increase purchase conversion rate?

It is best to focus on one problem for an operation. The refined operation for user retention can start from two aspects:

1) Data Model

There are three most commonly used data models for retention, namely the user funnel model, user life cycle model and RFM model.

The user funnel model is generally used to look at the overall situation, from traffic acquisition, to registration, to first purchase, to repeat purchase. The traffic is constantly decaying, and different problems may arise at each layer.

If there is too much loss from traffic acquisition to registered users, the problem may be that the traffic acquired is not accurate, and users leave after they find that the product is seriously deviated from the expected value;

If too many users churn between registration and first purchase, the problem may be that the new user guidance and new user benefits are not well done, and users are unable to quickly find the "satisfaction point" and complete the purchase;

If there is too much loss of users from first purchase to repeat purchase, the problem may be that the first-purchase discount is too great, attracting many non-target users to take advantage of the first-purchase discount. It may also be that the fulfillment of the first-purchase product is very poor, resulting in low user satisfaction and no repeat purchases.

The user life cycle model is generally divided into five stages: novice stage, growth stage, maturity stage, decline stage, and churn stage, covering the entire process from the user's contact with the product to the end of the relationship with the product. User retention is to extend the length of the growth and maturity stages as much as possible.

If the user volume in the growth and mature stages is an upward curve, it means that the current user retention status of the product is relatively healthy. If the user volume in the novice period accounts for too large a proportion, it means that the product cannot retain users and needs to continuously introduce new traffic to survive. Such a product will have problems sooner or later.

The RFM model is mainly used to monitor user activity and is interspersed throughout the entire user life cycle. Taking e-commerce as an example, by continuously paying attention to the user's consumption frequency, amount, and time, we can determine the user's current activity and value, stratify users according to RFM data, and then respond with different strategies.

2) Operational Strategy

The operational strategy for retention can be divided into three steps: identify the problem ➡️ find out the cause ➡️ formulate a strategy

The positioning problem depends on the data model mentioned above. 90% of the analysis without data support is nonsense. Don’t think that what you think is what you think. No one can be absolutely correct based on physical sensation.

When formulating the initial product operation strategy, you cannot understand everyone’s thoughts, so it is normal to have deviations. You can only rely on data to correct it step by step.

After analyzing the data reports, in 99% of cases the cause of the problem is already in front of you. For example, if you find that the order volume of a certain activity on the platform suddenly decreased by 30% yesterday, then you can list the three indicators of DAU, visit UV, and purchase conversion rate and do a month-on-month comparison. You may find that yesterday's visit UV dropped significantly. At this time, the cause is found: there is a problem with the exposure of the activity, then you can formulate a strategy around increasing exposure.

When formulating strategies, we must also start from the end in mind, first understand the user's expected value and the value that the product can provide, and then optimize the process or add new operational methods.

4. Seek breakthroughs

In the process of user retention, we often face this situation: it feels like we have exhausted all our weapons, but users are still leaving, and the churn rate is getting higher and higher, making you exhausted.

If you have encountered such a problem, you can stop and think about the following questions:

• What are the core needs of users?
• Are there any other demands that can be extended around the core needs?
• How can I exceed my users' expectations?

The reason for thinking about these questions is that no matter whether you are doing online or offline, it is difficult to have boundaries in today's business. The industry ecology has changed from incremental to stock, and all companies are competing for traffic. You can only ensure that your user pool is not penetrated by constantly meeting users' new needs and constantly expanding the boundaries of services.

After all, the best defense is offense.

Author: Shi Guipeng

Source: Shi Guipeng's Notes