Low cost and high activity, how to recall users?

What I will talk about today is mainly related to user operations . I hope that my sharing will be of some inspiration to your specific work, and I also hope that you can remember some key words .

1. How to decompose “ operational functions”

First of all, the first word is decomposition. How to solve KPI by decomposition? This KPI is about how to increase user activity at low cost.

So how to achieve the target? First of all, you need to know what the operation is doing.

1. Do a good job in content operation . Content operation is a matter of output and traffic , and the requirements for different websites and different products are also different. The output mainly comes from star companies and popular companies, while the traffic is mainly counted by users' views, reposts and interactions.

2. User operation. User operations mainly focus on user activity, which includes two parts: new additions and retention .

3. Product operation . Product operations are more concerned with product indicators, such as how many people use the product, how many conversions are generated, and how much project revenue. For example, Honor of Kings offers discounted heroes and also provides hero experience cards. When the product manager is doing optimization, there are one-day hero experience cards, as well as three-day and seven-day ones. When delivering content to users, operations should work together with product partners.

4. New operation here refers to new media operation . Taking the WeChat platform as an example, what we do is to increase the number of views of the content, which is a question of output and traffic. The number of new fans generated after browsing and the number of conversions that lead to purchasing behavior.

2. How to break down “active users”

“Active users” include new users and user retention .

1. New users

Where do new users come from? There are many, such as PC, app and new media . But for operators, mastering channels for user operations is secondary. The key is to master the method - the funnel model . How much traffic does the platform generate, how many clicks does it bring, and how many of them are new user registrations? Through optimization here, we can increase user click rate and registration rate , and ultimately reduce costs.

2. User Retention

User retention, including recalled users and old users. To recall users, we need to consider why and how they came. Some old users may return to the platform on their own initiative. For example, during the Spring Festival travel rush, they use 12306 to grab train tickets, but after entering the verification code, they find that there are no tickets available, so they have to come back the next day. Why are most products not that strong? How to recall the user at this time? This is today’s topic, low-cost user recall.

3. How to break down the recall process

1. Why do users come?

Here I suggest doing attribution analysis.

The first is the urgent need for job hunting . There are two types of such users. One is actively seeking a job, and the data shows that they log in frequently, submit applications frequently, and browse frequently in a short period of time. Occasionally, applications are submitted, which means that the applicant is preparing to change jobs, but only part of the requirements are met for this position. Both types of users are relatively popular.

The second is workplace knowledge. You may need some job-hunting skills, such as how to write a resume and how to communicate with HR. The browsing behavior of this type of users is mainly student users, and this is their need. Another requirement is professional skills, which is self-improvement.

The third is workplace information. It is mainly divided into three parts. One is salary analysis, such as the comparison between the salary of operations positions in first-tier cities and that in second- and third-tier cities. The second is the analysis of demand dynamics, such as the talent dynamics in each city. The third is career counseling, one is assessment, to understand how the position is evaluated before taking the position, and transition counseling.

2. How do users come?

It's nothing more than inside and outside the site. On-site messaging, EDM, push , etc., and off-site advertising effects and user sharing.

In fact, when users register, they will be recalled as old users. For example, in SEM, we can sometimes achieve a recall of 1:3, which is a side effect of registration. There is also user sharing, and when the content is particularly good, it will produce the side effect of sharing. There are also activities, which may be stimulating or red envelope activities. Of course, these activities involve expenses, so you can focus on reaching out within the site first.

So, how to reach users? It is necessary to segment the user’s reach methods and do scenario design.

Design scenarios are divided into existing methods and new scenarios. Currently, HR invites you to submit your resume and you can actively submit your resume for feedback. In addition to these two, there are four new scenarios: the first is job recommendation, for users who are actively looking for jobs; the second is job headlines, for users who are preparing to change jobs; the third is HR review, for users with market dynamics needs; and the fourth is career strategies , for users looking for job-hunting skills.

(1) Job recommendation - basic ideas

After adding new scenes, you will have a basic idea. For example, there is a cycle for user recall, from user submission - submitting job - collecting job positions and willingness - platform algorithm weighting - matching users and job recommendations.

(2) Job recommendation - basic process

When doing the process, the first thing you need to understand is that the basic idea is to have job-seeking behavior first, and then do user filtering. To match users who already have intentions, it is necessary to collect the users' intentions, match them with corresponding positions after collecting, and then form a list to recommend to the users, thus forming a user cycle.

What if the user has no behavior? According to the other three scenarios, match the scenarios for recall. By completing the user cycle in this way, the job recommendation process is basically completed.

(3) Job recommendation - execution process

Now that the process is ready, how do we execute it? On the first day, user data should be collected, including user application behavior, including job information and willingness. On the second day, we make algorithm recommendations and reach users through push, SMS, EDM, WeChat, etc. User follow-up is carried out on the third day, and users who have delivered goods continue to be matched by the algorithm to match other scenarios for the delivery users. Repeat the process on the fourth and third days, gradually cycling through the process and turning the recall into a repetitive process.

Job Recommendation - Manual Verification

When a platform develops a function, the technology often requires some data support. Where does this data come from? Manual verification is required.

• First, organize the algorithm. Present the user rules in written form, and formulate user screening rules and job matching rules.
• Second, extract the data. Cooperate with the data department to extract users who meet the rules within a period and match them with corresponding job information.
• Third, the copywriting reaches the target audience. The corresponding user's ID or contact information, as well as the user's corresponding position, need to be verified at this time. To verify the copy, you need to compare multiple sets of copy, and at the same time, do EDM and App outreach, and divide the users into ten groups for verification separately.

(4) Job recommendation - willingness collection

Users who have no intention of seeking employment can send a "direct line to famous companies". Since you have already registered, you only need to fill in the industry and company you are interested in, and your position, and the platform will be able to collect the user's willingness status.

(5) Job recommendation - landing page

Once the intention is expressed, the landing page of the job can be posted and the user will apply. The user opens the page and enters the job details to view the specific information, so that the manual verification is basically completed.

(6) Job recommendation - cross validation

After these are done, the effects need to be verified. From the graph, we can see that the entire data curve for sending texts 12345 is almost the same, which shows that these users have the same degree of recognition of the matching behavior. If there is overlap or imbalance, this data cannot be used as a reference.

(7) Job recommendation - verification effect

In effect verification, the big circle includes all target users , which is the overall goal. The second is reaching users. Reach refers to how much information is ultimately delivered to users. Neither EDM nor Push can achieve 100% reach. 86.4% of users receive reach messages, of which 26.1% are recalled and the last 19.5% are delivered users.

A quarter were recalled and a fifth were delivered. Through such verification data, we can prove to the product department whether this thing is feasible. Therefore, operations and products are inseparable, and operations need to interact well with products.

After all processes are implemented, the last step is to analyze the effect. After completing the KPI indicators, this process completely solves the problem of increasing user activity at low cost through recall.

Finally, all great things in the world must be done in detail, and all difficult things in the world must be done in easy steps. An operator has a big indicator for doing one thing. Under the premise of this big indicator, we need to continuously break down the indicator and refine it into clear and executable detailed indicators so that we know what KPI we are doing. Through continuous decomposition and optimization, it will ultimately help operators continuously improve their operational capabilities.

The author of this article @操作鹤手 is compiled and published by (Qinggua Media). Please indicate the author information and source when reprinting!

Product promotion services: APP promotion services, information flow advertising, advertising platform