How to carry out refined operations? See what Tencent's senior product manager said

When the product enters a relatively mature stage and has a large number of historical users or retained users, it is necessary to continue to attract new users. At this point, the user base is large enough that only one activity or operation strategy will definitely not be able to serve every user well. At this time, we need to consider differentiated operations , which are often referred to as refined operations.

1. What is refined operation?

Refined operation is an operational strategy that conducts differentiated and segmented operations targeting different groups of people, scenarios, and processes.

When is it necessary to carry out refined operations?

Generally, when a product enters a stable growth stage, it needs to rely on refined operations to increase user coverage in order to better serve users with different attributes and improve operational efficiency and effectiveness.

2. Three Thinking Angles of Refined Operation

Refined operations can be roughly divided into three aspects: people, scenarios and processes.

1. The refined thinking angle of the crowd

Users have various attributes, and each one is different. Therefore, we can refine the population by classifying them differently.

User flow model: new, retained, lost, returned

A user may be a new user or a retained user. Users may feel that your product is not good and become churned users, or they may come back and become returning users through some strategies after churn.

The user flow model is based on the flow of a single user identity. When a user has more than one identity, it can also be analyzed from other perspectives. For example, users can be divided based on their status: paying users or free users, relatively active users or users who need to be activated.

You can also think about it from the perspective of differences in user levels. For example, QQ members have different levels. Games are also divided into various levels, and these level differences can be used for grading.

There are also the users' natural attributes, such as age, gender, and region. Different operations can be performed for people in different regions and with different attributes.

Other identities defined in the product, such as whether you are a king or a diamond when playing Honor of Kings . Different identities have different demands for product functions and achievements. Using the same strategy for all these groups of people may affect operational efficiency, so refined operations can also be carried out from this perspective.

Case study of crowd-based operations:

Taking QQ membership service as an example, QQ members pay 10 yuan per month. In the early days, products that paid 10 yuan per month had a huge monthly user pool for the entire membership product.

New users come in every month and every day, but there are also customers who are not satisfied with the product or forget to pay and lose them. Some of the lost users are brought back through operational means and become returning users, thus forming a flow model for monthly users.

In the entire monthly subscription business, when the user scale gradually expands, for example, when the existing user base is 10 times the new user base, it is necessary to carry out detailed operations based on the user's life stage and mobility stage.

How to attract new users to pay? For example, free trial and 7-day QQ membership experience can be used to attract users to continue paying. You can also do a free trial or a low-price discounted trial afterwards. For example, one dollar for seven days of use. The payment threshold is set mainly to filter out those who cannot pay. The user's payment experience shows that the product is attractive to this user, so the conversion efficiency will be higher in the future.

In addition to considering the price dimension, you can also attract new members through the social dimension, invite friends to experience it through social invitations, and share the advantages and privileges of becoming a member with friends. Social invitations are also a very high conversion channel .

The third strategy to attract new users is to intercept payment points, such as QQ chat history roaming. By becoming a member, you can roam more chat records. In the scenarios and entrances for using the functions, it is stated that only members can enjoy the function, and there is a quick activation entrance so that users can obtain this function as quickly as possible, thereby attracting new people.

So once a person has become your user, how do you make him think that it is worthwhile to recharge his membership? Whether it is membership or other monthly subscription services of Tencent, many people are based on content, and the continuous updating of these contents is very attractive to users. QQ Show or QQ Dress-up, everyone has their own preferences, so the content must be continuously updated.

In addition to content, privileges also need to be constantly updated so that users can feel that their recharge privileges are constantly increasing in value, so that they can become retained users. At the same time, QQ membership is displayed through various logos and marks, which can satisfy the sense of dignity of paying users, which is also done in terms of retention.

What should we do for lost users? Some users churn because they forget to renew their subscriptions. In this case, we will send reminder messages to users seven days, three days, or one day before the expiration date. Although frequent reach is complained by many users, when conducting user surveys, many users said that the reason for loss was simply that they forgot to pay, so reach is very important.

At the same time, through Tencent’s social relationships, users are reminded that they have left the circle of member friends, thereby enhancing user stickiness and collective identity, making users willing to continue paying. This is also a form of reflux. In addition, for Tencent's privileged products, customers may leave if they lose interest in existing privileges, but may return if they encounter new privileges.

What if some users feel that paying is not worth it? Can be retained by discounts. The discounts are only available to certain groups of people and are not promoted publicly. Sending separate messages or email notifications is to ensure that the strategies targeted at different groups of people are only effective for this user.

2. Scenario-specific thinking

In addition to the crowd, in terms of the refinement of the scene, users may belong to different times, different locations, and use different functions. Based on these differences, there will be some differences in usage efficiency and conversion rate .

Case study of scenario-based operation:

There is an APP that advertises on WeChat , using two advertising positions : "bottom of public account articles" and "bottom of Tencent News articles." A daily budget of three thousand can bring about five hundred downloads. The customer thinks the download cost is a bit high and hopes to further reduce the download cost and improve the ROI of the campaign.

So, the 3,000 yuan is in two unknowns. The first is, where is the money spent and are there differences in the costs of each location? The second is when is it spent, and are there differences in the cost of different times? And what materials should be used to optimize the material space?

In this example, we performed location and time analysis. Before the APP was optimized, 3,000 yuan was spent every day on the bottom of the official account articles, and about 500 yuan was spent on Tencent News. There was a significant difference in the cost of a single download between the two. Then if we reverse the two budgets, we can take advantage of Tencent News' low cost.

Let’s look at the time again. This APP runs advertisements at two times, one is midnight and the other is eight o’clock in the morning. We will notice that the cost of a single download, this white line, the cost of downloading from zero to four o'clock is about six to nine yuan, and at three o'clock it is three to four yuan. Advertising from two to four o'clock is inefficient. So we let customers spend their budget during the day.

Put more budget into Tencent News and less into the bottom, and make good use of the cost advantage of Tencent News downloads to reduce the overall cost. At the same time, the budget is invested in the daytime. As you can see by the white line in the graph, there were 500 downloads per day before with 3,000 yuan, and now there are almost 800 downloads, an increase of about 60%. This is an example of refinement based on advertising placement.

Products and operational activities perform differently in different promotion locations and time periods. Try to analyze as much as possible where the product performs better and at what time so that you can perform more detailed operations. Therefore, doing things in the right scenario and at the right time can improve efficiency.

3. Process refinement thinking perspective

Process refinement is a case I worked on in the QQ Show team. QQ Show had a function at that time. If A changed QQ Show, he hoped to guide B to put on QQ Show as well. So what guidance should we provide? Let A switch to QQ Show to give gifts to three friends. The user must first select the gift-giving function, which will also send a notification to the friends. The friends click the notification and return to the activity page, which has a beautifully designed gift box. Click on the box to see A's gift, which can be worn for free.

The initial expectation for this function was to drive 100,000 people to switch to QQ Show every day. After going online, we found that 600,000 people exchanged QQ Shows every day, and each person could broadcast 3 QQ Shows, for a total of 1.8 million. However, only 15,000 people finally converted. So is there something wrong with the function?

We have conducted a detailed review of the process and found that 600,000 people use QQ Show every day. Without forcing users to enable the gift-giving function, only 200,000 users actually choose to use it. If each person sends three gifts, 400,000 notifications will arrive, 100,000 people will click on the notifications, 100,000 people will go to the page to see the beautiful boxes, only 30,000 will open the boxes, and in the end only 15,000 people will wear them.

How should this data be optimized? We recommend that when planning and operating a function, data points should be embedded in each process to understand the data status of each node, in which scenario users are lost, and which scenario transformation is more meaningful and valuable.

At that time, we had a limited number of developers and we didn’t want to force or harass users, so we simply removed the function of clicking on the gift box. As a result, the number of people wearing clothes became 40,000. This is done by segmenting and breaking down the process to clarify how to operate the product in a refined manner.

3. Implementation steps of refined operations

When we carry out refined operations in daily life, how do we usually implement them? First of all, there must be an operational goal, whether it is to increase revenue or increase activity and increase new users. Based on this goal, segmentation is done from the perspectives of population, scenarios and processes. Combine your own resources to make targeted operation plans , promotions and marketing .

During the operation process, we must constantly study whether the activities are in line with expectations and then adjust the plan. After the event is over, you can summarize and review it. Through summarizing and reviewing, you can know what can be improved so that the next event can be done better.

Points to note when implementing refined operations

Focus on the "main contradiction" and operate around the core goals. You need to know what the core problem is that this operation needs to solve and solve the most important problem first.

“Precision” should be applied in moderation and should not exceed the support capabilities of the team. The more detailed the population segmentation is, the better. Only when the population is significantly differentiated can the operation be differentiated. For example, the membership level churn rate. However, we do not do seven things for the seven levels every time. Instead, we operate separately for the levels with high churn rates. Different groups of people do different activities to achieve segmentation, and this segmentation is based on the team's support capabilities.

Focus closely on the target scene and group to avoid "leaking the secret". For example, discount activities for specific groups of people are not seen by old users, so as to avoid complaints from old users who pay full price. At the same time, for certain groups of people, they will think that this product is made with great care for them and their participation will be higher.

Try to use common strategies to create a word-of-mouth effect. It is best if the things you do for different groups of people have some commonalities so as to create a word-of-mouth effect. Whether it is WeChat or Alipay , for example, subsidies are provided for 711, with each transaction receiving a one-dollar subsidy, or two transactions per day with each receiving fifty cents. All scenarios are subsidies. Form a common strategy and reputation.

Automated and refined operations

The ultimate in refined operations is the realm of automated operations. First of all, there must be enough operational content to label each activity. The activity is aimed at new and paying users, and then the users are labeled. Then, a personalized recommendation system is used to display the operational content that best suits their label to the users.

A user of Toutiao has hundreds of tags, so he can find content from the corresponding tags and push it. The same is true for Guangdiantong . There are countless advertisements in the background and users have countless tags. Each time, the corresponding advertisements will be filtered according to the tags and pushed to you. This is the embodiment of automated and refined operations.