How to make the "user retention analysis model" more useful

Operations personnel often use user retention analysis models in their daily work to analyze user activity and other situations and find the reasons for fluctuations. At this time, a high-quality analysis model can bring a lot of help.

1. Project Background

1. What is the “User Retention Analysis Model”?

The user retention analysis model is an analysis model that analyzes user participation and activity. Through the model, operations/product colleagues can segment the scenarios that cause user churn, further identify user operation behaviors before churn and the causes of user churn, provide guidance for product optimization, and improve user retention rate.

2. Why do we need this iteration?

The user retention analysis model has been online on the Pangolin platform for nearly a year. During this period, the platform has been adjusted at the strategic level. In addition, during the use of the platform, users have reported problems such as "tabular data analysis is inconvenient, and retention analysis of users coming in through a unified channel cannot be performed", which led to this iteration.

2. Goals of this iteration

How can the retention analysis model better help product/operation colleagues analyze user retention? Combining platform strategy adjustments and user feedback, we set two goals for this iteration:

• Granularity of segmented retention analysis scenarios: Users can flexibly define the starting behavior, retention behavior, and target user groups for retention based on actual scenarios. Refined granularity can provide more valuable data for further analysis.
• Add comparison dimensions to table data - Add comparison dimensions for the two core indicators of retained users and retention rate. Multi-dimensional data comparison can help users make comprehensive comparisons, making it easier to analyze problems and improve analysis efficiency.

3. Competitive product exploration

In the selection of competitive products, we used industry reputation and functional completeness as evaluation indicators, and selected "Sensing", "GrowingIO" and "iResearch Ark" with excellent performance in all aspects from many analysis platforms as the objects of this competitive product analysis.

We mainly analyze this competitor from two levels. The first level is the user retention analysis model and content layer framework layout analysis.

We disassemble the content layer framework layout of each competitor's user retention analysis model one by one, as shown below:

By analyzing the disassembled content layer_framework layout, we can conclude that there is a general formula for the user retention analysis model: User retention analysis model = "retention rules" + "filtering conditions" + "table data display" + "visual data display" + "operation". This formula can be used in our iterative design plan.

In order to figure out the deeper meaning behind the existence of the formula, we interpreted the functions of the five functional modules (see the figure below) and found that we cannot delete any of the functions of the five functional modules, and the remaining functions will become incomplete after deletion.

The second level is the user retention analysis model field level analysis. We sorted out the model fields of competitors and summarized the fields of each module into a user retention analysis model field comparison table based on the functional module, as shown below:

By comparing the retention analysis fields, we found that some of the fields are shared by the three competing products, so we analyzed the shared fields (as shown below) and found that these shared fields laid the foundation for the user retention analysis model and are indispensable; the iterative design of the user retention model of our platform can be inspired by these shared fields.

4. Iterative Solution Design

The design of this iterative solution is mainly divided into three steps:

The first step is to determine the retention analysis function module and design the content layer framework layout.

Through competitive product exploration and analysis, we found that the general formula for the user retention analysis model is: User retention analysis model = "retention rules" + "filtering conditions" + "tabular data display" + "visual data display" + "operation".

Comparing the existing functional modules of the platform, we found that the "retention rules" module was missing. Considering the goal of this iteration - to increase the granularity of retention analysis scenarios, we decided to add the "retention rules" function.

Finally, we reconstructed the content layer of the retention analysis function, the framework layout, as shown below:

The second step is to determine the fields of each functional module.

In line with the idea of ​​running fast in small steps, daring to try and fail, and rapid iteration, we prioritize integrating the common fields of competitors and the existing fields of the platform. Focusing on the two major design goals of "increasing the granularity of retention analysis scenarios" and "enhancing the analysis and comparison dimensions of table data", the integrated fields are adjusted as follows:

The third step is to design a prototype of the retention analysis function.

In the specific prototype design stage, the designer combined the determined framework layout and module fields, focused on retention rules and table data display, and designed as follows:

Comparison of retention rules and table data modules before and after iteration:

1) Retention rules

Before the iteration, it was not possible to customize the starting behavior and retention behavior of retention. We could only analyze whether the users who logged in on the starting date logged in in the following days. Users could only select new and active users.

After iteration, customized starting behaviors, retention behaviors, and target users were added. In terms of starting or retention behavior, users can define any "event" as a starting or retention behavior, and can continue to add filtering conditions based on the "event" to refine the retention scenario; in terms of target users, the "user grouping" function data is connected with the target users. Users can select the created grouping in the "target user", or use the grouping function to create the retention scenario user grouping that needs to be analyzed.

2) Table data module

Before the iteration, the data display found that the data content was relatively small and the table utilization rate was low. Only comparative data could be formed vertically based on the retention analysis date dimension, and the analysis efficiency was low.

After iteration, data comparisons in three directions are added: vertical, horizontal, and diagonal. The vertical direction can compare the retention data of the same N days with multiple starting dates at the same time (see Figure 03 below), the horizontal direction can compare the retention data N days after the same starting date at the same time (see Figure 01 below), and the diagonal direction can compare the retention data of each starting date of the same date at the same time (see Figure 02 below).

5. User feedback after iteration

We conducted a user survey on our iterative solution by looking for users who had reported problems during the requirements collection phase and users who were currently using the platform. We obtained the following feedback:

• The display of table data has been optimized and enriched a lot, and it is more convenient to analyze. You don’t need to download the data first and then make the comparison table yourself as before.
• Customized target users in retention rules basically meet the needs of user analysis in specific scenarios. Previously, some retention analysis could not be done due to the lack of target user selection.
• It is very convenient to customize the starting behavior and retention behavior in the retention rules. Previously, we could only see whether the macro-level users logged into the APP. The more micro-level retention was not supported, and it was really painful not to get the corresponding data.

VI. Conclusion

This functional iteration is divided into four steps:

• The user feedback collected in the early stage combined with the adjustments at the platform strategic level determined the iteration goals;
• Analyze relevant competing products to derive general formulas and common fields for retention functions;
• Based on the conclusion of the competitive product score and the existing iterative design goals, the specific plan for this iteration was finalized;
• After the product was launched, we verified the iterative plan through user visits. Problems before the iteration, such as the inability to analyze user retention in multiple scenarios, the inability to conduct retention analysis based on specific user groups, and the difficulty in analyzing a single dimension of retention table data, were all resolved.

We will continue to pay attention to user feedback in the future. Some functions that are not implemented in the current version and problems exposed by users during subsequent use will be our future optimization direction.

I believe that with the correct strategic guidance, detailed competitive product analysis, rigorous design process, and scientific data verification, the user retention analysis model will become more and more perfect, the scenario data will be more comprehensive, and the user analysis efficiency will be higher.

Author: WOWdesign

Source: WOWdesign