Example Analysis | How do operators conduct App data analysis?

For an operations staff, doing a good job of data analysis can not only help us constantly adjust the direction of our work, but also prevent us from getting "deeper and deeper" in worthless mundane work.

One thing: every operation is very clear and very recognized. That is: data analysis is extremely, extremely, extremely...important to operational work!

But in reality, many operations are still "powerless". Their ideology has improved, but their actions cannot keep up. In other words, it’s not that they don’t want to do it, but that they lack a holistic understanding of data analysis, can’t find a clue, and don’t know where to start.

From the perspective of the effectiveness of the results, any work aimed at "completion" is not of much value, and any "subjective judgment" that lacks objective data indicators is not of much significance. This is true for the content production department that is responsible for process quality, and even more so for the operations and sales departments that are responsible for results.

For an operations staff, doing a good job of data analysis can not only help us constantly adjust the direction of our work, but also prevent us from getting "deeper and deeper" in worthless mundane work.

1. App-related data is divided into three categories

Data analysis is not something profound, let alone a technology. In the App operation work, the "data" we need to use can be generally divided into the following three categories:

1. Macro indicators: overall data such as total number of users, new users, retention , activity, and engagement;
2. User indicators: relevant data at the user level, which is the data support for all operational actions in the operation system;
3. Content indicators: Relevant data at the content level, which are the reference for improving and optimizing the platform content quality.

For the above three types of data, we only need to form "applied thinking" based on "full cognition" and achieve "flexible use" to easily handle any kind of data analysis work.

2. Macroeconomic indicator analysis

For an App, macro data includes five aspects: “total number of users, new users, retention, activity, and engagement.” Through these five indicators, we can fully understand the overall operating status of the App.

1. Total number of users: the sum of all users

(1) Data source : tracking point statistics

(2) Analysis focus: Filter out valid users from the overall App users and determine the number of truly valuable users. Generally speaking, an activated user who has completed registration and logged in to the device is a valid user; strictly speaking, a user whose App usage time > 0 is a valid user.

2. New users: a metric for promotion efforts

(1) Data source: tracking point statistics

(2) Analysis focus: During the App promotion process, many channels will adopt the method of “ H5 page registration + guiding App download”, which often causes the problem of some users “only registering but not downloading”. Therefore, for the number of new users, it is necessary to distinguish between "registration" and "activation".

3. User retention : a measure of user-content fit

(1) Data source: Umeng Statistics

(2) Analysis focus: This data is mainly used to monitor the retention of new users during a certain period of time (a day, a week, or a month). Judge user quality and App content quality through retention data.

4. User activity: the ultimate measure of App value

(1) Data source: Umeng Statistics

(2) Analysis focus: This indicator is the most core indicator among all App data and is also the ultimate standard for objectively measuring the value of an App. If we use one indicator to measure the value of an App, it is user activity data.

5. User engagement: a measure of user stickiness

(1) Data source: Umeng Statistics

(2) Analysis focus: This data is a measurement indicator of the depth of user App usage, including daily usage time, daily launch times, and other indicators. It is a useful supplement to the above four data.

summary

The above five indicators, total number of users and new users, need to be based on the statistical data of the technical department. As for the user retention, activity and engagement indicators, since technical statistics are more difficult, Umeng can be used as the standard.

Replenish:

(1) Tracking statistics: In order to ensure accurate tracking statistics, operations personnel and technical developers need to clearly define tracking data requirements and rules.

(2) Umeng Statistics: In order to ensure the accuracy of Umeng statistics, the following points should be noted:

• App Umeng SDK integration is correct
• The earlier the App is connected to Umeng, the better, to avoid some early data being unable to be queried
• Android and iOS versions are simultaneously accessible

3. User indicator analysis

The significance of user data analysis is to provide data support for all aspects of operational work, including new user acquisition , retention, activation, and payment conversion, from the user dimension.

For example, how to analyze the effectiveness of an App promotion campaign? How to adjust the content based on the operating habits of active users? Who are the paying users, and how long do they have to experience the app before paying?

Next, I will explain it to you in three steps: field decomposition, matrix analysis, and case practice.

1. Field decomposition

To analyze user indicators, you need to first break down the relevant fields of the user indicators. Taking the course app as an example, the user indicator data can be broken down into the following 8 fields:

(1) User ID

• register
• activation

(2) Mobile phone number

(3) Registration time

(4) Source channels

(5) Browsing behavior: UV, PV, Time

(6) Operational behavior (taking learning apps as an example): trial listening, subscription, recharge (time/amount/terminal), purchase (category/details), learning (audio/video), user completion rate (statistics by single section/statistics by single class), comment, course sharing, inviting friends

(7) Login behavior: number of active days, number of launches, and duration of use

(8) User Profile

• gender
• age
• area
• Profession
• Years of employment
• Education level
• Hobbies

2. Matrix Analysis

Step 1: We can regard these 8 fields as a horizontal matrix.

Step 2: Select one field (or one segment field) and export all eligible users

Step 3: Based on the exported users, we can analyze the characteristic data of these users in other dimensions to get the data we want.

3. Case practice

Example 1: We need to analyze the effect of a certain App promotion campaign to provide a reference for the next promotion campaign:

(1) Export all users brought by this promotion activity through the targeting field "Source channel";

(2) Through the "user ID" of this group of users, we can obtain the number of new users brought by the activity, the number of registered users, and the number of activated users;

(3) The quality of users brought by this activity is determined by the “user login behavior: number of active days, number of launches, and duration of use” of this group of users.

Example 2: We need to analyze the learning habits of active users to understand the reasons for their activeness and adjust the App content structure

(1) Export the top X users in terms of total active days by targeting the field “active days”;

(2) By comparing the data of the three indicators "UV, PV, and Time" of different pages in the " User Behavior -Browsing Behavior" of this group of users, we can find the access path that users are most accustomed to. Then we can judge whether the content display on the App homepage is reasonable, whether it can quickly arouse the user's interest and allow users to find suitable content, so as to make selective adjustments.

Example 3: We want to see who are the paying users and how long it takes for them to pay after using the app.

(1) Export paying users by locating the field “recharge amount”;

(2) Through the “user portrait” of this group of users, we can understand their gender, age, occupation and other information to provide a basis for advertising ;

(3) The difference between "each user's 'recharge time' - registration time" for this batch of users can be used to analyze the time it takes for users to consider payment from experience. Based on this, the content is adjusted in order to impress users more quickly.

4. Content Index Analysis

The significance of content data analysis is to comprehensively analyze user behavior in various dimensions from the perspective of content. Based on these user behaviors, we can provide a basis for content optimization and improvement.

Taking course content as an example, how do we judge whether our course text packaging and page display are attractive? How do we evaluate whether our courses meet user needs?

Next, I will explain it to you in three steps: field decomposition, matrix analysis, and case practice.

1. Field decomposition

To analyze content indicators, you need to first break down the relevant fields of the content indicators. Taking the course app as an example, the data of content indicators can be broken down into the following 10 fields:

(1) Course ID

(2) Course nature

• free
• Payment

(3) Online time

(4) Number of courses

(5) Course classification

(6) Lecturer

(7) Courses viewed

• UV
• PV
• Time

(8) Course subscription

(9) Audio and video playback

• UV
• PV
• Time

(10) Completion rate

• Calculate by single section
• Calculated per class

2. Matrix Analysis

The analysis methods for "content indicators" and "user indicators" are the same, both using matrix analysis. I won’t repeat it here.

3. Case practice

Example 1: How to judge whether our course text packaging and page display are attractive?

(1) Export the titles of all courses by locating the field "Course ID"

(2) By comparing the "number of subscribers" with the "number of viewed UVs", we can identify courses with higher subscription rates under the same display opportunities, and then determine which courses have better course packaging.

Example 2: How can we determine whether our courses have sales potential and then adjust the popular courses to display on the homepage?

To judge this problem, you only need to compare two metrics

(1) Comparison of “Subscribers” and “Viewed UVs”

(2) The value of the course completion rate indicator (calculated by each section or class)