Last time we talked about how, when locking in the cause of a problem, we need to use various data analysis methods for different problems to lock in the cause more quickly. We have previously talked about the [Funnel Analysis Method], and this time we will add several other methods and share them with you.

1. Comparative Analysis

The comparative analysis method is often used in life and work. It is also called comparative analysis method. It compares two or more interrelated indicator data, analyzes their changes, and understands the essential characteristics and development laws of things.

In data analysis, three categories are commonly used: time comparison, space comparison and standard comparison .

1. Time comparison

The most commonly used ones are year-on-year and month-on-month . By comparing data over time periods, we can understand the current data level.

Year-on-year comparison: comparison of a period in a certain cycle with the same period in the previous cycle, such as June this year compared with June last year, Monday this week compared with Monday last week, etc.

Month-on-month comparison: Compare a certain period of time with the previous period of equal length, such as this week compared with last week, etc.

When analyzing year-on-year and month-on-month comparisons, you need to pay attention to the scope of comparison, as well as when to use year-on-year comparisons and when to use month-on-month comparisons. For example, when measuring the effectiveness of an activity, the data of the previous activity and the current activity should be analyzed year-on-year.

When performing visual output and displaying year-on-year and month-on-month data, if you are not familiar with the year-on-month calculation, you can refer to the year-on-month function of BDP, and you can freely choose the time for comparison based on the data situation.

2. Spatial contrast

That is, to compare with different spatial indicator data within the same time range.

For example: by comparing different departments, different business personnel, different regions, etc., for example, by comparing the differences in order sales data among provinces, we can conclude that the product's advantageous regions should be the focus of breakthroughs and balance human and material resources.

3. Standard comparison

Business data usually sets target plans. Standard comparison can be used to understand the current development process, completion progress, etc. by comparing the current data with the set target plans. After understanding the gap, the strategy can be adjusted in time.

For example, set target values, average values, medians, and other standards in the chart to form a standard comparison with the actual data and analyze the data situation.

2. User Analysis Method

User analysis is the core of Internet operations. Commonly used analysis methods include: activity analysis, retention analysis, user segmentation, user profiling , etc. In the RARRA model just mentioned, user activity and retention are very important links. Through the analysis of user behavior data, the product or web design can be optimized, and users can be properly guided.

Usually we monitor daily user activity data such as "daily active users" and "monthly active users" to understand the new active user data and whether the product or webpage has attracted more attention. However, at the same time, we also need to do retention analysis to pay attention to whether the new users are actually retained and become regular users. Retention data is the real user growth data, which can reflect the usage of the product over a period of time.

The activity rate is the proportion of active users to the total number of users in a certain period of time. According to time, it can be divided into daily active rate (DAU) , weekly active rate (WAU) , monthly active rate (MAU) , etc.

But the definition of active users may vary from product to product. Some apps are considered active when they are opened, while others are considered active only when you log in...

Why is activity rate so important?

This goes without saying. "The conversion cost of a new customer is approximately 3-10 times the cost of an active customer" and "the 2-8 principle" all illustrate how important activity is. As an operator's important KPI, do you really know how to analyze it?

Scenario case: (one quarter is a life cycle)

User A downloaded the product and started using it. He found that it met all his needs and couldn't put it down. He logged in almost every week and the login time was more than 2 hours.

User B downloaded the product and started using it, but stopped using it after a few days. After the product was updated, he thought the new features were great and started using it again. After that, the frequency of use was about once every two weeks.

User C casually registered after searching online, used the product for a few days, and found it just average. When the product had a lot of discounts or promotions, he used it again one or twice. It was used less than 5 times in one quarter.

When there is a new user acquisition campaign, user D uninstalls or gives up using the app after downloading and registering, and the number of times he uses the app in the entire quarter is 0 or 1.

The above four types of users can be divided into the following categories according to their activity level:

Active users: (User A)

User activity path: New-Active-Loyal

Countermeasures: Ensure contact frequency, but do not provide promotional incentives

Silent user: (User B)

User active path: Added-Inactive-Return-Active

Countermeasures: Ensure contact frequency and provide a small marketing discount

Sleeping user: (User C)

User active path: Added - Inactive - Return

Countermeasures: Control limited contact and redeem through discounts

Users in the churn period: (User D)

User activity path: Added - Inactive - Lost

Corresponding measures: Block contact and only notify users during big promotions such as "Double Eleven". It is necessary to determine the operation and marketing plan based on the user's active path and promote the user's final conversion.

So how do we monitor activity rate, retention and other data to see if they are normal? We need to pay attention to several indicators of data changes:

1. Fluctuation range: whether there is a large fluctuation in a short period of time

2. Change persistence: Is data fluctuation persistent?

3. Change regularity: Is there a certain regularity in data changes?

4. Correlation of changes in various indicators: Is there a certain correlation between the changes in various indicators of concern, such as rising and falling at the same time, the same change trend, etc.

3. Segmentation Analysis

Today, when the concept of data analysis is widely valued, it is difficult to truly discover problems through rough data analysis, and refined data analysis has become a truly effective method. Therefore, the segmentation analysis method is to make the original data analysis more in-depth and refined.

For example, to analyze the curriculum conversion situation in Beijing, we need to subdivide it into different student types, students at different stages of primary, middle and high schools, different areas of Beijing, the different situations in Haidian, Chaoyang and Xicheng, different subjects, etc. In the process of data analysis, we go from coarse to fine, using rough data to show the overall situation, and then refine to the local and analyze the specific reasons.

The key point of segmentation analysis is to classify the data and analyze them separately by category.

There are two implementation methods:

1. Multi-layer drilling

Nest each layer of data, click on different dimension data to conduct segmentation analysis, and through multi-layer drilling, click directly in the chart to view segmented data. Each layer of data can select the appropriate chart type for display.

2. Focus on drilling down

Focus analysis is performed on some key data in the data. In the overall analysis, if you want to view the details of some data that you are particularly concerned about, you can use the focus and drill-down functions to perform free analysis.

4. Index Analysis Method

In actual work, this method is the most widely used. It is also a method of highlighting the key points of the problem while using other methods for analysis. It refers to the direct use of some basic indicators in statistics to do data analysis, such as mean, mode, median, maximum value, minimum value, etc. When choosing which basic indicator to use, it is necessary to consider the orientation of the results.

Average : It can show the data of the same type in different time periods and is used to summarize trends and find problems in general rules. In addition, you can also compare the differences in similar data in different regions and under different circumstances, which is more convincing than the total amount or individual value.

Median : also known as the middle value, refers to the number in the middle position of a set of data arranged in order. It represents a value in a sample, population or probability distribution, which can divide the set of values ​​into two equal parts. Because it is obtained through sorting, it is not affected by the maximum and minimum extreme values. For example, when calculating the recruitment salaries in the market this quarter, it is more meaningful to present them as the median, as there may be a small number of maximum or minimum values.

Changes in some data have no effect on the median. When individual data in a set of data changes greatly, it can often be used to describe the central trend of this set of data.

Maximum (minimum) value :

The maximum (minimum) value can often be used to show "abnormal" situations in the data. In some data analysis, outliers can be ignored, but some analysis of the maximum (minimum) values ​​can study the influencing factors, thereby finding breakthrough actions or avoidable methods to promote business growth.

5. Funnel Analysis

The funnel analysis model is an important method in business analysis, and is most commonly used in marketing analysis. Since each key node in the marketing process will affect the final result, in today's world where refined operations are widely used, the funnel analysis method can help us grasp the efficiency of each conversion node and thus optimize the entire business process.

Among them, we often focus on three key points:

First, what is the overall conversion efficiency from start to finish?

Second, what is the conversion rate for each step?

Third, which step has the most loss and what is the reason? What are the characteristics of lost users?

Funnel analysis usually helps us solve more than just conversion rate issues. Refined funnel analysis can also help us:

1. Funnel comparative analysis: finding optimization methods from differences. Funnel analysis that compares different user groups and different marketing methods can help us quickly discover user characteristics and conversion advantages of marketing methods, and find steps that can be optimized for different users in the conversion process, or areas that can be strengthened in marketing methods.

2. Use conversion rate to locate the most effective key method for conversion. Most of the commercial realization processes can be sorted out into funnels. Usually we will take a variety of methods to increase conversion. Funnel analysis can help us sort out the entire business process and identify the most important conversion nodes. Therefore, in the process of analysis, we can find out whether there are other unimportant processes involved, which affect the conversion of the main process, so as to make trade-offs and optimization.

Funnel model typical case

AARRR Analysis Model

Acquisition, Activation, Retention, Revenue, and Refer, namely user acquisition, user activation, user retention, user revenue, and user dissemination.

As can be seen from the figure, this is a typical funnel chart that decreases step by step. By analyzing the changes in the conversion rate of each link, we can determine the key method to achieve the final conversion and continuously optimize and iterate.

In the current situation where Internet products are generally in a red ocean, growth experts have also made new thoughts and optimizations on the model. In the AARRR model, the most important focus is on user acquisition, and the final conversion is improved by expanding the traffic pool at the top of the funnel. In the current market situation, customer acquisition has hardly become the most important indicator for growth. The redefined RARRA model helps people shift the focus from user acquisition to user retention, which requires more attention to user activity and retention data, which is also an important analysis indicator in the user analysis method we will talk about later.

Author: Little Strawberry?

Source: Little Strawberry?