With the development of the Internet, increasingly complex business logic, data analysis will become increasingly important. Analysis of the data logic can effectively avoid confusion, preventing the complex business logic clear understanding, misjudgment. Here we give ten common share data analyst data analysis ideas.
Taoism emphasized words, called "Road Law, Tactics and implement."
Levels are as follows:
"" Refers to objects or tools, referring to the field of data analysis tools is the product or data analysis, "we must first of its profits";
"Skill" refers to the technique, the level of skills, efficiency to compete, such as the use of technology tools for analysis;
"Act" means the method of choice, there is a saying "more important than trying to choose";
"Road" refers to the direction, it is the guiding ideology, is strategy.
Data analysis and products, operations optimization, data analysis method is its core, is "law" and "skill" level.
So how do the data analysis of it, today's top ten methods of data analysis we Laijiangjiang.
01 Segment analysis
Segmentation analysis is the basis of the data analysis, the index data under a single dimension information value is very low.
Segmentation can be divided into two categories, one is a step by step analysis, such as: Visitors to Beijing can be divided into Chaoyang, Haidian District; the other is the dimension of the cross, such as: SEM of new visitors from paid.
Subdivision for solve all the problems. For example, the conversion funnel, in fact, is to follow the steps in the conversion process is broken down, the flow channel analysis and assessment also requires a lot of use subdivision.
02 Comparative analysis
Comparative analysis mainly refers to two interrelated indicator data to compare, display and explain size of the study, the level of height, speed, speed, etc. relative value of quantity, by comparing the index under the same dimensions can be found, to find out at different stages of the business problem.
Common methods include comparison: comparative time, spatial contrast, a standard comparison.
There are three time comparison: up, the chain, than the fixed base.
For example: This week and last week is to compare the chain; and the first week of last month year on year comparison is the first week of this month; all data contrast ratio was fixed base with the first week of the year. There are three ways you can analyze information on the level of business growth and speed.
03 Funnel Analytics
Conversion funnel analysis is the basic business model analysis, the most common is the final conversion to achieve a particular purpose, the most typical is to complete the transaction. But it can also be achieved for any other purpose, such as the first use app for more than 10 minutes.
Funnel help us solve two problems:
If a leak occurs in the process, if there is a leak, we can see in the funnel, and can be further analyzed to plug the leak by.
In a process if there are other processes it should not occur, resulting in the conversion of the main process receives damage.
04 synchronization group analysis
Cohort (cohort) data analysis is very important in operational areas, particularly Internet operators need to be carefully retained insight into the situation. By the nature of exactly the same comparable comparison groups on retirement, to analyze what factors affect user retention.
Cohort analysis is an important reason for the popularity is very simple, but very intuitive. Cohort with only a simple graph directly describes the user in a period of time (or even the entire LTV) retention or loss of changes.
Analysis of previous retained as long as the user has a return visit is defined as the retention, which can lead to artificially high retention index.
05 Cluster Analysis
Cluster analysis has a simple, intuitive features, the site analysis is divided into clusters: the user, the page or content source.
A user clustering mainly grouping users, user tagging; page clustering are mainly similar to the relevant page group method; sources including channel clustering, key words.
For example: the page analysis, there is often a page with parameters. For example: Information details page, product pages, etc., all belong to the same category page. Simple analysis is likely to cause bounce rate, exit rate and other indicators is not accurate, it can obtain accurate data for analyzing the same page scene by cluster analysis.
06 AB test
One of a growing hacker's main idea is not to do things a large and comprehensive, but continue to make small but things can quickly verify. Quick verification, how to verify that it? The main method is the AB test.
For example: You find the conversion funnel in the middle of a loophole, it must be assumed that the problem of commodity prices led to a loss, you see the problem - funnel, also came up with the idea - to change the pricing. But the idea is correct, depending on the real user response, so the use of AB testing, some users still see the old price, the new price is part of the user to see if your idea really work, the new price should have a better conversion, should this happen, the new price should be finalized, and so forth optimization.
07 Buried analysis
Only data collected sufficient basis to get the required results by various analytical methods.
By analyzing user behavior, and broken down into: browsing behavior, interaction with mild, moderate and severe interaction, transaction behavior, for browsing behavior and interactions with mild events such as the click of a button, because of its frequent use, data simple, no-Buried technology buried self-realization, which can improve the effectiveness of data analysis, data needed to be extracted immediately, but also reduce the workload of a large number of technical personnel, the need to gather more information-rich behavior.
Such as: severe interaction (registration, invite your friends, etc.) and transaction events (plus cart, orders, etc.) are buried way point of the SDK is implemented by batch.
08 Source Analysis
Traffic dividend disappear, we attach importance to the degree of high-acquisition sources, how effective source tagging users, is essential.
Traditional analytical tools, channel analysis only a single dimension, to in-depth analysis of the sources of the effects of different channels at different stages, SEM and paid search and other cross-analysis in your area, received a customer come in different regions of the details, the finer dimensions, results the more valuable.
09 user analysis
User analysis is the core Internet operations, commonly used analysis methods include: active analysis, retention analysis, user grouping, user portrait, scrutiny and other users.
Users can browse active subdivided into dynamic and interactive activity, trading activity, etc., through the active behavior of the subdivision, master key behavioral indicators; be grouped by user behavior sequence of events, user attributes, observed clustering of users to access, browse, register, interactive behavior, transactions, etc., in order to truly grasp the characteristics of different types of users, to provide targeted products and services.
User portrait based labeling system will automatically complete user portrait drawing clearer, more effective decision-support operations.
Analysis of Form 10
Fill out the form is an essential part of each platform and user interaction, excellent form design, to enhance the conversion rate plays an important role.
From the user enters the form page, it creates micro funnel and successfully submit the form from the total number of persons entering the final completion, in this process, how many people are beginning to fill out a form, filling out the form, can not cause any difficulties encountered complete the form, all affect the final conversion effect.
These are the common methods of data analysis, application method requires more flexible application based on business scenarios.