Some time ago, I wrote a series of data analysis articles on the Internet industry to analyze specific business scenarios.
Today brings you a large screen of Internet traffic monitoring. Without any code base, you can teach everyone to quickly and efficiently build an enterprise-class data management cockpit in 10 minutes. Purely dry goods, if you can bring some help in data visualization analysis, will not be very pleased.
The data analysis tool used in this article is FineBI, a visual BI analysis tool that can be used for both enterprise-level big data analysis and personal data analysis. Here we need everyone to start with the next FineBI tool (linked at the end of the article), connect the data and perform the following steps.
A business package
Before constructing the data visualization analysis, first familiarize ourselves with the underlying business data model table that we need to analyze (FineBI has built-in Internet industry business packages, which have data tables). The related data tables involved in this data visualization analysis are mainly five tables: regional dimension table, promotion channel dimension table, user dimension table, access statistics fact table, and access phase statistics fact table. Here, these can be directly used by FineBI tools. Data table passed key Associated and modeled automatically. In this way, when the front-end analysis is performed, the dimensional indicators between the tables are linked, and the multi-dimensional perspective analysis can be rolled up and down in any angle.
Area dimension table
The regional dimension table mainly contains the location information of the region and can be used to associate with the access statistical fact table to obtain specific regional data.
User Information Dimension Table
The user information dimension table mainly contains relevant personal information of the user, such as the user name, age, and gender, and may be used to correlate data with the access statistical fact table and the fact table of the access statistics phase to obtain basic user related information.
Promotion Channel Dimension Table
The promotion channel dimension table mainly includes three levels of promotion channel names, which can be used to access the statistical fact table to correlate to obtain the source information of the promotion channel for the visiting user, so as to facilitate the user’s return source tracking and improvement of the promotion channel.
Access statistics fact table
The access statistics fact table mainly includes the information such as the user’s staying time, pageviews, visits, and the number of times of enumeration under the corresponding statistical date, and may be related to the promotion channel dimension table, the user information dimension table, and the regional dimension table to obtain correlations. Dimension information.
Access phase statistics fact table
The access phase statistics fact table mainly contains the total residence time of the platform for each access phase of the user, and may be associated with the user information dimension table to obtain the related information of the access user.
Second, the regional users view statistics
First of all, for the Internet industry’s enterprises, its users are often all over the country. For example, we set up a network server in Nanjing, and its platform access users may come from all over the country. An analysis system that includes data such as geographic information locations like this, using a built-in data map is more appropriate.
Next we want to count the distribution of users in each region. As you can see, the original data of the map dimension table in the business package only contains three fields: the area, the area ID, and the parent ID. If we need to make map data statistics according to the province-city region hierarchy as shown in the final rendering, how can we achieve this?
FineB has a function called a self-looping column that can hierarchically display hierarchical data, such as time (year-month-day) and territory (province-city-area/county), similar to “14 March 2018 “Day” extracts fields of “2018”, “March” and “14th.” As shown in the following figure, two-tier data can be layered on the region dimension table, the region ID is selected in the hierarchical ID column, the parent ID is selected in the parent ID column, and then the self-recursive column is directly constructed to store the data in the regional table. You can.
The following figure shows the result of layering from the circular column.
After obtaining the provincial-city data with hierarchical relationships, we dragged the level 1 and level 2 of the regional dimension table to the map component classification box, and the fate fields are the province and the city. The access statistic facts are dragged into the indicator box. The page views in the table. Then beautify it, layer the values by color, and range from 0 to 500, 500 to 1000, 1000 to 2000, 2000 to 3000, and 3000 to unlimited intervals. Then you can see the different provinces, the distribution of the pageview (color depth represents the size), and then click on a province, you can see the corresponding data from the drill to the city for user browsing data analysis and statistics.
Third, the key indicators statistics
For key information viewed by users such as page views and visit times, we can use the FineBI Dashboard to collect statistics on direct data indicators and visually display the progress of key indicators for each platform. As shown in the figure below, select the number of visits, number of bounces, and number of visits fields in the access statistics fact table, drag it to the indicator box of the dashboard, and select the ticker type dashboard in the style attribute.
After that, they can directly see the statistical results of their indicators, which are 23917, 8267, and 9175 respectively. The number of jump outs is not much different from the number of visits, and the rate of missed jumps is relatively high, indicating that the content of the platform needs to be optimized.
Fourth, the type of user loss rate analysis
Next, we hope to conduct statistics on visit miss rate data among different user-level groups, and observe and analyze the differences in the number of visits, jump-outs, and miss-loss rates between different user groups.
First, we select the user type field in the user dimension information table as the classification axis of our composite graph. Then the left value axis drags and selects the number of visits and the number of times to jump out. The right value axis adds the new calculation index of the missing rate = the number of bounces / visits . At the same time, the chart type of the indicator, the left-axis indicators are set as a histogram, the right-axis indicator is set as a line chart, and the unit order of magnitude is modified to a percentage format for the analysis of the data chart.
As shown in the above figure, the VIP users, old users, and new users of the platform have a missed rate of 97.15%, 91.00%, and 87.37% for the three types of user platforms. This seriously indicates that the users of the platform are not sticky enough. Well, most users are based on one-time consumption and fewer repeat customers. For this type of Internet platform, you may consider doing more platform product promotion activities for old users and VIP users. At the same time, do more research on old users and return visits of old users, and enhance the interaction between the platform and users, and improve users. And stickiness between platforms.
Fifth, the user visits to visit TOP10
Another very important analysis is user activity statistics. Taking the user’s platform view statistics as an example, we often pay more attention to the more active users of the platform and then carry out targeted incentives or users to create new operations. First, we select the user name in the user information dimension table as the classification axis, and then the value axis places the index of visits in the access statistical fact table to count the total pageviews generated by each user on the platform. Next, we filter the user name field. The filter condition is set to the first 10 users in the pageview. Then, the pageviews are sorted in descending order according to the page views.
It can be seen that the three user platforms Blanche, Henry, and Christina have very high page views. In the top three, the platform needs to focus on active user operations, for example, adopting some new incentive systems to generate new users on the platform. Drainage effect, while achieving a win-win situation between the platform and the user.
Sixth, platform users age traffic statistics
For the Internet platform, users are our food and clothing parents, so the statistics on the features of users’ portraits have guiding significance for our channel promotion. Here we use the platform’s user age segment traffic statistics, first drag and drop the age field in the user information dimension table (in the built-in business package, the age field of the user information dimension table is a string type, where we can first add at the ETL Columns are converted to numeric types) into the classification axis, and then the view volume field in the access statistic fact table is selected as the statistical indicator in the left value axis.
At this point, we will find that FineBI automatically performs group statistics with a step size of 10 according to the user’s age, that is, according to 10-20, 20-30, 30-40, and 40-50, the platform access interval statistics for all age groups. . From the data in the above figure, it can be seen that the platform’s main user groups are young people between the ages of 20 and 30, which is also in line with the age group distribution of contemporary Internet users. In view of this, the platform can conduct targeted marketing for young users ranging from 20 to 30 years old when doing some channel online or offline marketing promotion.
VII. Distribution of page views by channels
For the promotion of platform drainage, we usually need to conduct targeted promotion of various channels to achieve the effect of new user drainage. Therefore, we need to monitor the effectiveness of each marketing channel in a timely manner. The promotion channel of the platform is divided into three levels. For this type of multi-level data analysis and statistics, selecting a multi-layered pie chart for statistical calculation is more appropriate. We drag the first-level channel name, second-level channel name, and third-level channel name in the promotion channel dimension table to the classification axis in the multi-layer pie chart component, and then the index box contains the number of visits in the access statistics fact table. In the field, enter the style property setting interface and adjust the inner diameter of the multi-layer pie chart to 60.
As shown in the above figure, most of the platform’s current pageview promotion channel sources are new media marketing channels, occupying 64.59%, but user-generated platform views generated by offline sources are only 14.43, which has much room for improvement. . In addition, we can easily view other promotion and drainage effects of other channels through multi-layered pie charts and monitor them in time to find out and solve problems early.
Eight, the user behavior phase statistics radar chart
Finally, let’s take a look at the analysis of the residence time of the users of the platform during each phase of the behavior to understand how much time the user spent in each phase. For example, if you browse the product for too long, you can consider whether the platform’s search engine and product recommendation mechanism are perfect; if the payment stage draft cost is longer, whether it can be because the platform’s payment channels are not rich enough, you can consider adding some more generic other payment channels. Wait.
Like the statistical analysis of this type of long and short board comparison data, we can often use radar charts to display the relevant data. First, we classify the axis to select the final access phase field in the access phase statistics fact table. At the same time, the series axis selects the access platform field, the index axis selects the total residence time field, and then the new calculation index is added to convert the original total residence time unit to the original For seconds, divide by 3600).
As shown in the above figure, we have successfully counted the time spent by users in all stages of their activities. It is not difficult to find the overall time spent by users visiting the platform in each phase. IOS > Android > Mobile Browser IOS users still occupy the mainstream group for the platform. In addition, the time taken by users to place orders and add shopping carts is not too short. This section can consider whether platform users have room for optimization in the process of placing orders and adding shopping carts.
Nine, layout and color
The above is the analysis modules that I hope to display on the Internet data screen, and they will be built into an Internet cockpit for data monitoring. The built-in data here may not be scientific, and the analysis may not be rigorous enough. The main purpose is to make everyone familiar with the production process and specific operations.
After each analysis module is produced, we then add some beautifying elements to the entire dashboard, such as title, background, and some animation effects, to enhance the overall delicateness of the dashboard.
Here I set the dashboard background (R,G,B) to (0,0,0), the title bar background (R,G,B) to (36,38,64), other colors refer to the global brought by FineB Style color can be.
Finally, it can beautified by repeatedly previewing and adjusting the gaps, sizes, and proportions between components.
The above is the process of making large screen data. Some operations may not be detailed enough. You can use the product help documentation (link at the end of the article) to enter the door.
The core of the big screen is still the data analysis of each module. For themselves (data analysts, business analysts), a big screen is like a report that tells a leader a story and emphasizes completeness. Second, the leaders looked at this large screen and were able to see at a glance the data they were concerned about. They knew that the indicator was low and where the problem was. This also achieves the purpose of supporting decision-making.
Finally, I hope that the young partners can quickly get started, analyze the business needs according to the actual data of the company, and create a similar management cockpit for their own boss. The promotion salary increase is just around the corner!