How to build your user system?

The past and present of the user system

This title is a bit broad, but what we are going to talk about is actually quite simple.

Let us first go back to the United States after the Civil War. After the end of the Civil War, the U.S. economy was in decline. It is said that in 1793, an American businessman designed a very simple reward system for consumers. Consumers could get a certain number of copper coins for each purchase. As long as consumers kept shopping in the store, they could accumulate copper coins. When a certain number was accumulated, they could redeem gifts in the store. This practice was adopted by many other merchants in the 19th and 20th centuries to acquire long-term consumer customers.

This may be the earliest known user system. This system simply and roughly divides users into three types:

• Potential customers: People who have not purchased yet but may purchase
• Buyers: Users who have made purchases
• Frequent customers: users who frequently make purchases

In the early days, the user system of Internet companies did not really exist. Because early Internet companies simply divided users into registered users and visitors, because at that time most Internet companies did not have the need for refined operations . The Internet field at that time could be roughly divided into information sites and game sites (later, PC games). When the profits of most Internet companies were based on advertising revenue after the traffic volume increased, whether or not a company had a user system was not the key factor that determined the failure or success of the company's overall operations. Instead, whether there was enough traffic to obtain enough cash flow or revenue was the key factor in whether the company could survive or even live well enough.

So, when did the user system begin to be taken seriously?

My opinion is value-added services and PC games.

Early PC games charged by time, so as long as the player played long enough, he would spend more money than others. After Shanda started charging for props, players first recharged tokens with cash and then consumed the tokens in the game. At this time, the original length of time that determined the stickiness of old users has become the recharge amount or consumption amount that determines the user's stickiness. At the same time, value-added services have caused differentiation among users who originally used the product for free, with some of them providing more value to the product by purchasing the value-added services.

As a result, the membership systems that already existed in traditional industries (airlines, credit cards) began to enter the Internet.

Moreover, the original intention of entering is very consistent: companies should provide more rights and services to users who give them more value, thereby rewarding user loyalty and forming a positive cycle.

But with the development of the Internet, this system has also been constantly changing, from focusing on loyalty to returning to focusing on value itself, which has promoted the formation of a more complex user system today.

This is probably the past and present life of the user system.

Tiered incentives for users

In "Starting Operations from Scratch", I emphasized user segmentation and gave the RFM model . User segmentation is actually something that most companies do. As for why we need to segment, we actually talked about it in yesterday's content. You can see it here: How to understand "Operations make product suggestions in reverse"? User life cycle is one of the reasons. Another reason is that for companies that can measure user contributions, it is an inevitable choice to divide users into different levels and give more rewards to users who provide higher direct contributions.

This is also why Taobao launched super membership this year and JD.com modified the terms of its membership system to replace the original membership tiers with Jingxiang Value, because companies hope to give benefits to users who have made direct contributions to them, rather than wasting them on users who contributed before but now do not contribute or contribute little.

It is foreseeable that more companies will join this approach in the future. Of course, the safest way is to purchase a membership service, just like Brother Liang chose Amazon 's Prime membership.

For transaction-based companies, defining user contribution is actually a simple and crude three-dimensional process: proximity, frequency, and amount. However, for non-transaction-based companies, this approach is difficult to implement because the contribution value is not so easy to measure simply without transactions. But, I will still do it.

At this point, you will see:

1. Tool products choose advanced account versions other than free and paid versions
2. Service-oriented products reduce costs and maintain users’ original service usage behavior, including the cooperative monthly card model of Mobike and ofo to pass on costs
3. User Referral is a cost-controlled customer acquisition method that is becoming more and more popular

…

There is actually no fixed method for user segmentation. No one can really tell you how to segment your users, because each company’s business is different, which is very embarrassing.

The purpose of stratified goals is actually to motivate.

• Encourage high-value users to continue to provide high value;
• Encourage the second-highest value users to contribute higher value;
• Clean up worthless users to save resources.

There are many ways to motivate, and the user system is the epitome of them.

Because what the user system needs to do is:

• Discrimination against users
• Resource Tilt
• Incentive Interference

And it is done in a routine.

Basic methods for building a user system

Although no one can tell you how to do your own user segmentation, there are actually some basic routine methods for building a user system. Let me briefly talk about them.

Step 1: Sort out the performance of user contributions

The first step, of course, is to simply sort out the business, identify your valuable users, and then bury data at the points where users will contribute; the next step is to summarize the contribution level of user behavior .

For example, for a content platform, the key is, of course, to match your content consumers with content contributors who are your contributing users. Then, to measure their contribution, it may be their level of activity (including number of logins, interactions, and content creation), content quantity, and content quality. For a supply and demand platform, the most critical thing is, of course, the level of activity on both the supply and demand sides and the total supply and demand provided. This needs to be sorted out according to your own business.

Step 2: Define the Stake Pool

If you want to build a user system, "discrimination" is bound to occur, because the platform resources you can provide are limited, so you must provide more resources to high-contribution users in exchange for their contributions. Therefore, the equity pool must be a priority consideration, including how to gather resources, how to allocate these resources, to whom to allocate them, and to what extent. These are all things that need to be paid attention to, and they must be considered from the very beginning. In principle, the equity pool can only be expanded, not reduced. Of course, it can be replaced.

Step 3: Classification

Users must be "discriminated against". Without discrimination, there would be no way to divide users into different levels and carry out targeted operations. The classification of users must be done very carefully. No matter what method you use to classify the users, the RFM-based model can be used as a reference and adjustment object.

It should be noted that no matter what kind of hierarchical model you have established, you must know that this matter is not over yet, you must continue to move forward, and the following steps are critical.

Step 4: Run data and test run

After the model is built, you need to run it in the background for a period of time based on real data. It is recommended to run it for at least 3 months. The reason for running it for such a long time is to see the distribution status of users based on their real actions, and what your resource costs will look like in conjunction with such distribution status.

The standard for official launch is very simple:

It is indeed possible to allocate resources in a hierarchical manner to users who make contributions while keeping costs under control.

That’s about it. Give it a try and see what specific problems you want to solve.