App Data Embedding Classification method

1, the importance of data buried point

In the real work, the overall flow of data is: data Production-collection-data processing-analysis and mining-data visualization , which, the acquisition is a very important part of the data collection is not complete, right, directly determine the breadth and quality of data, the impact of all subsequent links. If the collected data dimension is very small, then we want to conduct in-depth analysis of the time will appear to be nowhere to start, such as we want to analyze the use of a certain feature of an app, but there is no data collection, we can not carry out subsequent analysis. If data acquisition is wrong, such as false negatives, false positives, and so on, then this is certainly very deadly, and will mislead our data-based decisions.

In the Internet industry, the way of data collection according to the different data collection, mainly divided into Web data collection, app data collection. The collection of Web data is mainly using JS acquisition, commonly used data analysis tools are mainly Google Analytics,app collection is mainly through the buried point, the main front end of the buried point and the point of burial points, the corresponding mobile data analysis vendors are also many. As more and more users tend to surf the web and use App,app data acquisition, it becomes more important.

2. What is the data burial point?

To do a good job of data analysis, so that data support decision-making, products and operations, the first is to have data, like cooking requires rice, and to collect user data will be the application data buried point. Product personnel in the product planning is to think about the problem of data buried, if the product after the outside to consider how to bury the point, it will lead to the previous version of the user data can not be collected, want to see a data when there is no alternative, only wait until the new version perfect to make up. When thinking about which points to bury, it is a brainstorming process to tightly combine products, operational requirements, and communicate with data analysis, ETL personnel, etc., such as which metrics need to be monitored, what buried points need to be implemented, and how to bury points more in line with statistical requirements.

The data burying point includes the client front end burying point and the service end burying point, the client front end burying point is the user on the client's operation record, the service end is the user online request server log, each has the merits and demerits, the best both are buried points, can complement each other.

Data buried point is actually in the product function code after the addition of statistical Code, record user action behavior, when the user to do something, such as click on a button, rendering a page, etc., the development will be in the corresponding place with statistical code, buried point will generate a log record.

3, how to bury the data point

3.1. Data burying Technology

Now the industry client buried point is generally taken key-value form, key represents an event, value represents the corresponding parameter value, the buried point logic is probably: User generated click Behavior, generate a click event, when there is a page rendered in front of the screen, generate a page event.

Now there are two main types of SDK escalation incidents: 1) Click events, mainly describe the user's click behavior within the app, such as Click tab, click Button, and some parameter events are also categorized as click events, such as page description, audition length, etc. Classify these parameter events as click events is mainly convenient for page events to calculate the length of the user's application, 2) page events, mainly describes the user visited pages, such as home page, details page, etc., while the length of time spent on the page to calculate the length of the user application.

There are three common techniques for burying the dots:

Code embedding : When a control operation occurs, the data is sent by pre-written code. Advantages: Control the time of sending data, detail record of event custom attribute, disadvantage: time, labor cost is big, data transmission timeliness.

Visual Embedding : Using visual interaction, through the visual interface to configure control operations and event operations, through the background screenshot of the way to collect data. Advantages: Low cost, fast speed, disadvantage: Less behavior record information, less support analysis method.

No burying point : When the user presents the interface element, the event is triggered by the control binding, and the system has an interface to allow the developer to handle the behavior. Now there are two main non-embedding practices in the market, one is to pre-track all the rendering information, one is the lag tracking rendering information. Advantages: No need to bury the point, convenient and fast, disadvantage: less information on the behavior record, high transmission pressure.

3.2. Advantages and disadvantages of data burying point

Advantages and disadvantages of 3.2.1 and client burying points

The benefits of the app client embedding point are:

1) be able to collect more comprehensive user data, can the client each click, the page buried point;

2) You can collect data that does not require a request server, such as local playback of music, length of page stay, etc.

There are some problems with the app client data burying point:

1) data reporting needs mobile phone networking, if the user's mobile phone is not networked, it will delay the escalation of data, while the data escalation generally have the corresponding rules, will not be reported in real time, there are generally about 15% delayed escalation and false negatives;

2) If the user deletes their own app action record, there is no reported record.

3) When the need to change the buried point, you need to update the version of the line, but there will be some users do not update the version, affecting the quality of data.

The advantages and disadvantages of 3.2.2 and service end burying point

The benefits of App server backend embedding are:

1) Real-time collection, data is accurate, there is no delay escalation;

2) When changing the buried point, as long as the change, the report data will change;

3) ability to collect actions that do not occur within the app, as long as the server is requested, and the client can only collect operational behavior in the client, such as counting the amount of installs that are drained from other apps.

The disadvantages of the end-of-service backend are:

1) data that does not require a request server cannot be collected;

2) Users are not able to collect data when they are not connected.

3.2.3, server and client integration

Combined with the advantages and disadvantages of the end-point and end-of-service end of the client, the service-side data is high real-time and accurate, the best users need to request the server's critical business volume are used server backend buried points, such as online playback, game installation, in the statistics of the winning users, obviously the service data more reasonable, Client data may miss some of the winning users, resulting in user complaints; The client data is very full, the majority of user operation behavior is recorded, other non-critical traffic or do not need to request the server's behavior after using the client front end buried point. Service end burying point and client buried point each have merits and demerits, should two kinds of data at the same time exist, can mutually confirm, when one party data big problem can be found through the other side, at the same time, data can be complementary, such as the acquisition of sudden problems, you can use the other side data replacement.

3.3. Data buried Content

The data buried point is mainly to collect the user behavior data, according to the user application behavior, each application burying point mainly includes following several aspects:

3.3.1 , Function button burying point

The buried point of the function button is to understand the user's use of buttons, functions, portals, etc. in the application, so as to gain insight into User Content preference, function preference, use path and so on. Mainly include the following aspects:

1) The user to the Application Top tab, the Bottom tab, the Content Filter tab and so on the click, to understand the user to the application each plate preferences and content preference;

2) users to the application of the entrance, channel Click, to understand the user's content preferences, product use path, etc.;

3) Users of the application of the specific content of the click, such as Click on a game, a product, a video and so on, to understand User content preferences, product use path, etc.;

4) The user clicks on the options in the app settings, such as whether to open the push switch, to understand the basic settings of the user application.

5) The user clicks on other function buttons, such as the pause in the music player and the click of the Next button.

3.3.2 , main behavior buried point

The main behavior of user application is to understand user's application behavior, so as to gain insight into User Content preferences, product usage habits, user loyalty, user activity, etc., the main behaviors of user application include:

1) User core behavior, including download, installation, audition, playback, reading, etc., each business according to the nature of their products, there are different core business behavior, such as video playback, music audition, you can understand the user's content preferences, business activity, etc.

2) User other behavior, including comments, sharing, likes, replies, collections, subscriptions, etc., the content-based business is generally common, you can understand the user's content preferences, usage habits, business loyalty, etc.;

3.3. 3 , Main Page burying Point

Page embedding is to understand the user's view of the application of each page, so as to know the number of page views, user Use path, length of use and so on. Mainly includes the application homepage, the Personal Center page, each level page, each two level page and so on, the principle is that as long as the application renders in front of the screen needs the corresponding buried point, so as to be more accurate calculation application use time, simultaneously can evaluate each page value.

3.3. 4 , payment process buried point

The burial point of the payment process is to understand whether the entire payment process is used smoothly, whether there is room for optimization, easy to funnel analysis, the application of the payment process is critical. This includes steps such as Click to pay, select payment method, click Next, confirm payment (or cancel payment), pay results, etc. to understand the conversion rate of each step and optimize the problem payment link.

3.3. 5 , search process burying point

Search is a very important feature of the application, most of the targeted users will use the search function, search for the buried point to understand the user's search behavior, content preferences, search hotspots and so on. Mainly include click to search, enter keywords, click recommended hot words, click on the recommended Lenovo keyword, click on search results, etc., insight into how users search, the current search hotspot is what, search recommendation algorithm effect and so on.

App Data Embedding Classification method