Website data analysis: User segmentation-based comparative analysis zz

From the user level of the website, we divide users into different types based on the behavior characteristics of user access, because the user behavior is different, the behavior statistical indicators are different, and the analysis angle is different, therefore, if you want to classify users in detail, you can implement different classifications based on various rules from many perspectives. We have seen that some data analysis reports have made various user segmentation and analysis of various user behaviors, combined with other dimensions, it seems that the content is rich enough, but it is difficult to understand what the analysis results are to explain, it may be appropriate to use a consulting report to reflect the current overall trend and user characteristics, but if we really want the data analysis results to guide us to do something, we still need to determine the purpose of analysis before performing user segmentation, and clarify the requirements at the business layer.

Since we need to make a comparative analysis based on user segmentation, we naturally want to clarify the differences between the behavioral characteristics of some user groups and other user groups. This article focuses on the adjustment of the guiding content level as the guide. By comparing the differences in content requirements of different user segments, We can optimize content operations, recommend high-quality content or content that meets user preferences to corresponding users.

Since it is based on user segmentation, we first define the user segmentation rules. Here, we will give an example of three types of subdivisions: Lost and retained users, new and old users, single-purchase users, and secondary purchase users, based on these three categories of subdivisions, we conduct comparative analysis on the purchased products of each category, and identify which products meet the user's expectations.

　Comparison between lost users and existing users

Of course, to distinguish between lost users and reserved users, you must first define user churn clearly, for details about the definition of lost users, refer to previous blog articles-active users and lost users of websites. With the definition, we can make statistics and segments, or take e-commerce websites as an example. The content of e-commerce websites is commodities, we calculate the loss of users who purchase these items based on each item, as shown in the following example:

　　

The indicator definition here should be clear. The proportion of lost users of each product should be the proportion of the number of lost users after the product is purchased among all users who have bought the product, however, we only know that the loss of users of each product cannot be compared to the example to determine whether the product promotes user retention, or to some extent, the loss of users is caused, only by comparing with the overall level can we draw a corresponding conclusion. Therefore, it is important to explain how the value "compare with the overall" is calculated. The percentage here is not a result of direct subtraction, but a manifestation of the difference, assume that the overall user traffic loss rate is 56%, take product A as an example. The comparison result is as follows: (58.13%-56%)/56% = 3.80%, the same calculation method can be used to obtain the difference between other products and the overall comparison. Finally, it is displayed. In Excel, the results in the plot can be directly displayed through the data bar function in the "condition format", which is very convenient.

Obviously, the analysis results shown in the figure above provide direct guidance for Operation Adjustment to promote user retention, so what we need to do is to recommend the products that are beneficial to the user (the user turnover rate of F products is much lower than the total, indicating that F products are more conducive to user retention) to the user, and optimize or remove the products (C Products) that may lead to user loss.

　Comparison between new and old users

Similarly, the above method can be used to differentiate the purchase bias of different user groups. The segmentation of New and Old users is the most common method of user segmentation. We can use a similar method to see different product preferences of New and Old users:

　　

What do you see from this? The proportion of new users who buy D products is obviously low. Maybe new users do not like this product at all, while B Products and F products are obviously more in line with the taste of new users. If your website supports targeted promotion by new and old users, the above analysis results will benefit you a lot.

Of course, the features presented by this data may have a certain relationship with the promotion channels of commodities. For example, many products of D use centralized promotion channels (such as EDM) of old users ), naturally, the proportion of old users will be high; or some products will be displayed on the Landing Page where new users are concentrated, the proportion of new users who buy the product will obviously be high. Therefore, when doing such analysis, you need to pay attention to the differences in promotion channels and the specific analysis of specific problems, which cannot be generalized.

Comparison between single-purchase users and second-purchase users

The same method can also facilitate multiple purchases. For e-commerce websites, the user's first shopping experience is very important, which will directly affect whether the user will purchase again or after multiple times, or whether it can become a loyal customer of the website. If your website focuses on user relationship management, you can try the following analysis method:

　　

It should be noted that the basic user group here is set for the first-time users (not all) of each product. We want to analyze all the cases where this product is used as the first-time Purchase Product, whether the user will initiate another or even multiple purchases, so as to evaluate the effect of the product on the first purchase experience. From the table above, we can see that product B and product F are not doing well in Secondary Purchase. It is very likely that the use or quality of the product affects the user's satisfaction and hinders the user's re-purchase. Based on the analysis results, we need to pay special attention to products with a much lower secondary purchase rate than the overall level, and also analyze the product features, some products are more likely to facilitate secondary purchases, because there may be cross-selling and upward marketing situations.

In fact, I originally wanted to split this article into multiple topics into a series of topics, because in terms of implementation, the analysis of each segment of user segmentation needs to be completed independently, and most of them need to be calculated from the underlying data. If you look for similar data from Google Analytics, the only thing you can find is the new access ratio, in addition, the metrics subdivided into each page in the content module do not contain % New Visits (this metric is included in the traffic source and region segmentation ), of course, you can customize a report to view the new access ratio of each page of the website. The benchmark is the overall new access ratio of the website, the GA display method selection directly provides the "Comparison" view compared with the overall view, which is my custom report:

　　

The effects displayed on GA are similar to those after the condition format is customized on Excel 2010 (2010 shows the red and green data records with positive and negative values on the left and right sides of the coordinate axis, 2007 seems to have not yet implemented this function). This benchmark-based comparison presentation is very intuitive and can be used in other analyses. So what do you see in the comparative analysis of the proportions of new users in my blog? The obvious trend in the first few articles is that the ratio of new users in the conceptual methodology is higher than the average value (of course, mainly relying on the help of search engines ), the proportion of new users in opinion and analytical articles is lower than the average value (old users prefer practice and application). Therefore, if my blog can dynamically display different content to new users and old users, this analysis will be very valuable. Maybe your website can try it.

Finally, back to the problem at the beginning, we need to sum up that the subdivision is used for comparison, and the comparison is to reflect the differences and make adjustments and optimizations. Therefore, the goal of subdivision is to ultimately guide operational decision-making, this is the value of data analysis.

This article uses the BY-NC-SA protocol, reproduced please note the Source: website data analysis» user segmentation-based comparative analysis