Buried data often heard the word after work, but do not understand what Buried Yes. Reference answer about God's great to know the answer to almost
Know almost Original: https: //www.zhihu.com/question/36411025
The first answer
Author: bullhead
link: HTTPS: // zhuanlan.zhihu.com/p/25 195 217
Source: know almost
copyrighted by the author. Commercial reprint please contact the author authorized, non-commercial reprint please indicate the source.
The point is to collect so-called buried in the application specific process some of the information used to track the status of applications used, subsequent to further optimize the product or to provide data to support operations, including the number of visits (Visits), the number of visitors (Visitor), stay duration (time On Site), number of page views (page Views) and the bounce rate (bounce rate). Such information gathering can be roughly divided into two types: a page Statistics (track this virtual page view), statistical operations behavior (track this button by an event).
Buried in the way of data
Now the mainstream Buried in two ways:
- The first: their own R & D statistics inject code in their products, and to build up the appropriate background check.
- The second: third-party statistical tools, such as Friends of the Union, Shence, Talkingdata, GrowingIO and so on.
If you are early products, often use the second approach to data collection and analysis directly using third-party tools for basic analysis. For those more concerned about data security, business and relatively sophisticated companies usually use the first method to collect data and build the appropriate data products to achieve the demands of the application or its data analysis.
key indicator
Let's take a look at both the APP, H5 or applet indicators are concerned, to understand the nuances and the complexity of the calculation method of these indicators, another angle to think about the meaning buried point. [Source: proficient Web Analytics 2.0 ]
- Access and Visitors
Visits (Visits) and the number of visitors (Vistors) is almost all applications require statistical indicators, which is the most basic indicators.
For statistical applications, it is often seen DAU, MAU, UV and other indicators refer to statistics visitor (Vistors). Access (Visits) refers to the session layer, the user opens the application to spend some time browsing and leave from the definition of indicators (visits) for this session is called statistics (Session) number.
A session (Session or Visit) is a request to open a first application (open application) request and a final decision. If a user opens the application and then put down the phone or away from your computer, and no action within the next 30 minutes, the session automatically ends, usually counted as one visit or session period (30 minutes is the application of the convention sessions early Web version definition, long a user stays longer present at the time of application, a 30-minute limit will also be different, in short, is to represent the length of time a user accesses).
In calculating the number of visitors (Vistors), Buried reported data is as close to the real number of visitors. There is a need for statistical indicators of this scene unique visitors, there is still need to emphasize that the number of visitors (Vistors) is not a real independent person, and therefore must know when to collect data access Although the number of visitors can well reflect the true use of the application the number, but not equal to the number of people using real applications. (The reason is that repeated application installation, or phone parameters are modified will make the index unique visitors affected. Buried calculate the number of access is dependent on Cookie, the user opens the application, the application will be created in the person of a separate terminal Cookie, Cookie will be retained, but still inevitably be manually clean up the Cookie is disabled or lead users to use the same application Cookie inconsistent, so the only unique visitors a height close to the real number of people using the application.)
- Long-stay
Long used to measure a particular page in the application or user visit time (session) the residence time of stay.
When the page to stay long, it represents the time spent on each page; for example: First is to enter the Home (10:00) to leave the home page to the next page (10:01) The long, long calculated as 1 minute to stay home. A page is 2 minutes. Long data collection is not always necessarily get the residence, such as page B to enter the time (10:03), leaving abnormal or exit time is not recorded, this time is calculated 0 (so you need to know when buried point index calculation condition , excluding such invalid data).
How long they stay in applications, visit time (session) of the stay, it is a long calculation time access to all pages, and also on a process, which is 4 minutes long stay application.
- Bounce Rate
Calculation of bounce rate is still very present in many various companies, the most frequently used are: visited only a proportion of the session page share (reason: Suppose this scenario, users access a page on the left think about the user's mind the picture should be: open the application, I thought what the hell, then close the application even uninstall this dreadful scene, which is why so much attention is the bounce rate index)
Bounce rate can be broken down into two levels: first, the bounce rate of the entire application, the second is the focus of the landing page bounce rate, and even search keywords bounce rate. Index bounce rate operability is very strong, you can find the page in question found the problem directly through keyword statistics bounce rate.
- Exit Rate
Exit rate for the page is the goal of this indicator is very simple, that is, how many users left the application for a page, the main users reflect the situation from the user application to leave. Which pages need to be improved is the fastest way to explore. ( Note: The exit rate is not necessarily a bad thing. For example: exit of the final node of the forecasting process should be high)
- Conversion rates
We have invested so much in the product, not just to measure outputs it? So for the electricity business class applications, as well as more worthy of attention right index than the conversion rate? Conversion rate is calculated by dividing the output of some kind of unique visitors or visits, the electricity supplier for the product, the user is submitting orders divided by unique visitors.
Calculate the conversion rate seems to think that simple, but it is buried in the closest service point of data collection. This is also reflected in most indicators Buried skills, it requires a combination of business features developed calculation method. Submit orders / number of visitors is the most basic conversion rate, conversion rate can be hierarchical, specifies the path to the user, such as: the completion of a path submit orders / number of visitors.
Try to find a path, think about how data conversion come of it, buried point what kind of data collected, right?
- Participation
Engagement is not an indicator, but collectively referred to a series of indicators, such as access to the depth, access frequency, the number of orders for the electricity supplier, the serial number of content providers for these players, and user behavior can be a measure of engagement index of. The reason why the participation as an indicator, the indicator is hope you understand the business combination, a chemical reaction, learn and use to discover the essence of things.
Buried in content
After reading these key indicators, in fact, buried point roughly divided into two parts, one is the application of statistics page visits that page statistics to be reported with the action occurs when the page is accessed; another part is operating within the statistical behavior of the application, in a page when the operation to be reported (for example: the component exposed when the assembly click on the slider, when falling).
For statistical indicators to need, all the pages in the application, the events are unique tag information, user information, equipment, time parameters and meet the business needs of the specific content parameters are appended to report, is buried point.
Notes about the data points buried
- Do not over-pursuit of perfection
Buried on data it is crucial, Buried in order to better use the data, do not try to get accurate data to get a high-quality point data buried, bounce rate data is discussed in this example, been able to get in front of , ever use the US data to reach the next move, not the pursuit of high-quality precision. This is a lot easier data products into the pit of the earth, we should always remind ourselves.
The second answer
link: https: //www.zhihu.com/question/36411025/answer/468092622
Source: know almost
copyrighted by the author. Commercial reprint please contact the author authorized, non-commercial reprint please indicate the source.
Production Data - Data Acquisition - Data Processing - Data Mining and Analysis - The data driver / user feedback - product optimization / iteration.
Buried: The first step in data analysis
Big Data, from behind complex data mining, analysis of user habits and preferences, identify a more user "taste" of goods and services, combined with targeted user needs to adjust and optimize its own, it is the value of big data . The collection of this information, analyze it not open around the "Buried."
Small science: "Buried"
Buried point is collected in a position need the corresponding information, just as the camera on the road, can be collected to the property of the vehicle, such as: color, license plate number, vehicle information may also be acquired behavior of the vehicle, such as: There is no red light, there is no pressure line, the speed number, the driver has not answered the phone during driving, etc., if the camera distribution is the ideal state, then by superimposing the different positions of the camera acquired, can restore the path of a car, destination, even infer the driver's driving habits, whether it is old driver information.
So, every point is like a buried camera, capture user behavior data, the data is multi-dimensional cross-analysis, the user can restore the real scene, and tap user demand, thereby enhancing the user maximum value of the whole life cycle.
4 kinds Buried "posture"
In order to collect huge amounts of data to be more accurate, as the follow-up to create a "pure" data analysis environment, Buried technology came into being. Consolidate data base or not, depending on how the data is collected. Buried variety of ways, according to Buried different locations can be divided into the front end (client) and the rear Buried (server) Buried, wherein the distal end Buried comprising: a code Buried, whole Buried visualization Buried .
Buried point : The SDK, to collect operational data pages all controls, the "statistics screen" configuration characteristic data to be processed.
Advantages: all operations are buried point, simple, fast, no statistical data on-demand processing Buried
Disadvantages: Upload consumption data flow, data dimensions single (only click, load, refresh); affect the user experience - user process prone Caton, seriously affecting the user experience; more noise, data accuracy is not high, easy interference; can not customize buried collect information.
As satellite photography, one by one without mounting the camera, but a huge amount of data, and is easy to miss, difficult to dig the key information, and therefore all buried point manner, for the following scenarios:
For example scenarios ·
Mainly used in simple pages, such as: short-term activities in the landing page / topic page, click on the need to quickly measure the effectiveness of distribution.
Buried JS Visualization : embedding SDK, circle defined visual event
For convenience products and operations students can be a simple circle directly on the page to behavior (defined events) to track users,
Only collect click (click) operation, saving development time, Zhuge io was recently supported by JS visual Buried.
Advantages: The interface of the configuration, not to develop, convenient Buried update to take effect soon
Disadvantages: Buried custom properties to support the poor; need to reconfigure the page when remodeling or changes;
As satellite aerial, without having to install a camera, a small amount of data, support information acquiring local region, so JS visualization Buried more applicable to the following scenarios:
For example scenarios ·
1, fast track data collection methods: Active / H5 and other simple pages, business people can directly circle, no threshold operation, reduce intervention and technical personnel (from World Peace), such data collection methods to facilitate operational as soon as possible to master pages conversions key nodes, but lighter application of user behavior data and can not support more in-depth analysis;
2, if the page temporary adjustments, additional Buried flexible, can be used as supplemental code buried point, increase in time to collect data
Buried Code : embedded SDK, custom events and add event code, on-demand collection, business information better, more focused analysis of the data, so the code Buried is an analytical business value for the starting behavior.
Advantage: comprehensive and accurate data collection, to facilitate subsequent in-depth analysis (Buried accuracy of the order: Code Buried> Buried Visualization> Buried point), SDK is small, no effect on the application itself experience
Disadvantages: need to coordinate R & D personnel, there is a certain amount of work
For example scenarios ·
1, if you do not want to collect data at the same time, reduce the user experience
2, if you do not want to collect vast amounts of useless data
3, if you want to collect data: finer granularity, more dimensions, higher accuracy of the data analysis
Well, consider the long-term value growth, please select code Buried
Buried service end : to support other business data acquisition and integration, such as CRM and other user data, through interface calls, the structured data, as collected directly from the server side, higher data accuracy for their customers with the ability to collect or may be combined acquisition and acquisition client.
For example scenarios ·
1, by calling the API interface, the CRM data such as user behavior data integration, the full amount of multi-user perspective;
2, if the company has its own buried point system, it can be collected directly from the server to upload user behavior data to Zhuge io platform for data analysis, maintenance buried two point system;
3, open up the historical data (data before burying points) and new data (Buried after), improve data accuracy. As access client after the client acquisition, after introduction of the original historical data, previous existing user access to the platform it will not be marked as a new user, reducing data errors.
How buried?
Buried sounds "Li Jue unknown", is actually very simple, like "cameras installed on the road."
1, carding product user behavior to determine the distribution event
Buried program distribution scheme camera mounted ≈
There are children's shoes often consulting Zhuge Jun: exactly what data acquisition to data analysis? Answer this question, first clear purpose, to clarify the logic.
Zhuge io objects and basic data analysis of user behavior, select the recording and analysis of user behavior which directly affect the value of output analysis work, Zhuge Jun Recommendation: Select the product target and current user behavior most closely related to the first problem, as event. With electricity providers, for example, the behavior of each user process is defined as a type of event and gain distribution of logical events.
2, recording events, understand and analyze user behavior
Information to be recorded to determine the camera is illegal to take pictures or speed?
After recording and analyzing user behavior needs to sort out, and complete event distribution table, then, need the assistance of the R & D engineers, according to the type of platform (iOS, Android, JS) your application complete access to the SDK, each distribution event will become a very brief program code - when you do the appropriate behavior, your application will run this code, the corresponding event record to Zhuge io. In the distribution is completed, when the hair product version, users start using the new version of the application, usage behavior data will be automatically transferred to Zhuge io, so you can make the following analysis.
This step, Zhuge io's CS team will provide support for enterprises, assist the technical team successfully completed the first step of data collection.
3, by recording identify the user identity
Zhuge io recorded in the user's behavior, namely: what the user did? In the process of user analysis, there is one type of information is useful, namely: who the user is (TA's id, name) and have what characteristics (age, TA, types ......)? You can identify process Zhuge io platform, it will pass the user's identity and characteristics to Zhuge io, the use of information identify the fine analysis:
细分用户群：用户属性的一个很重要的作用就是将用户分群。您可以根据identify的属性定义筛选条件，进行用户群的细分，比如用「性别=女」的条件将所有的妹子筛选出来，然后分析妹子们的行为特点和转化率……
基于属性的对比：细分的重要目的之一就是对比，您可以基于「性别」细分，然后对比「妹子们」和「汉子们」的行为、转化、留存等的区别；
基于属性的人群画像：您可以基于用户属性，对产品的任意用户群进行「画像分析」——该用户群的男女比例、地区分布、年龄层次、用户类型……
最理想的埋点方式？
回到一开始的问题：何种埋点方式最理想呢？
正如同硬币有两面，任何单一的埋点方式都存在优点与缺点，企图通过简单粗暴的几行代码/一次部署、甚至牺牲用户体验的埋点方式，都不是企业所期望的。要满足精细化、精准化的数据分析需求，可根据实际需要的分析场景，选择一种或多种组合的采集方式，毕竟采集全量数据不是目的，实现有效的数据分析，从数据中找到关键决策信息实现增长才是重中之重。
因此，数据采集只是数据分析的第一步，数据分析的目的是洞察用户行为，挖掘用户价值，进而促进业务增长，故最理想的埋点方案是根据根据不同的业务和场景以及行业特性和自身实际需求，将埋点通过优劣互补方式进行组合，比如：
1、代码埋点+全埋点：在需要对落地页进行整体点击分析时，细节位置逐一埋点的工作量相对较大，且在频繁优化调整落地页时，更新埋点的工作量更加不容小觑，但复杂的页面存在着全埋点不能采集的死角，因此，可将代码埋点作为辅助，将用户核心行为进行采集，从而实现精准的可交叉的用户行为分析；
2、代码埋点+服务端埋点：以电商平台为例， 用户在支付环节，由于中途会跳转到第三方支付平台，是否支付成功需要通过服务器中的交易数据来验证，此时可通过代码埋点和服务端埋点相结合的方式，提升数据的准确性；
3、代码埋点+可视化埋点：因代码埋点的工作量大，可通过核心事件代码埋点，可视化埋点用于追加和补充的方式采集数据。