Ten commonly used data analysis methods
In-depth analysis of big data Taoism emphasizes four words, called "dao, magic, magic, and device".
Level difference:
"Apparatus" refers to objects or tools. In the field of data analysis, it refers to products or tools of data analysis. "Workers must first sharpen their tools if they want to do their jobs well";
"Surgery" refers to the operation technology, which is the level of skill and efficiency, such as the technology used in analysis tools (such as the level of data analysis using Excel);
"Law" refers to the method of choice. There is a saying that "choice is more important than effort";
"Tao" refers to direction, guiding ideology, and strategy.
In terms of data analysis and product and operation optimization, data analysis methods are its core, belonging to the level of "law" and "technique".
So how to do a good job in data analysis, today we will talk about the top ten data analysis methods in Internet operations.
Practicing the great power of data analysis is not an overnight effort, but continuous growth and sublimation in practice. A good data analyst should be value-oriented, look at the overall situation, based on business, be kind to people, and use data to drive growth.
01 Segmentation analysis
Segmentation analysis is the basis of analysis, and the information value of indicator data under a single dimension is very low.
The subdivision method can be divided into two categories, one is a step-by-step analysis, for example: visitors to Beijing can be divided into Chaoyang, Haidian and other districts; the other is dimensional intersection, such as: new visitors from paid SEM.
Segmentation is used to solve all problems.
For example, funnel conversion is actually subdividing the conversion process according to steps. The analysis and evaluation of traffic channels also require a large number of subdivision methods.
02 Comparative analysis
Comparative analysis mainly refers to the comparison of two interrelated indicator data, showing and explaining the relative values of the research object's scale, level, speed, etc. in terms of quantity. Through the comparison of indicators in the same dimension, you can find and find out Business issues at different stages.
Common comparison methods include: time comparison, space comparison, and standard comparison.
There are three kinds of time comparison: year-on-year, ring-on-month, and fixed-base ratio.
For example: The comparison between this week and the previous week is the month-on-month comparison; the comparison between the first week of this month and the first week of last month is the year-on-year comparison; the comparison of all data with the first week of this year is a fixed base ratio. There are three ways to analyze information such as business growth level and speed.
03 Funnel analysis
Conversion funnel analysis is the basic model of business analysis. The most common is to set the final conversion to the realization of a certain purpose, and the most typical is to complete the transaction. But it can also be the realization of any other purpose, such as using the app for more than 10 minutes.
The funnel helps us solve two problems:
Whether there is a leak in a process, if there is a leak, we can see it in the funnel, and we can stop the leak through further analysis.
Whether there are other processes that shouldn't occur in a process, causing damage to the main transformation process.
04 Synchronous group analysis
Cohort analysis is very important in the field of data operations, and Internet operations particularly require careful insight into retention. By comparing the retention of comparable groups of exactly the same nature, we can analyze which factors affect the retention of users.
The important reason why cohort analysis is so popular is that it is very simple, but very intuitive. The cohort only uses a simple chart to directly describe the retention or loss of users over a period of time (or even the entire LTV).
In the previous retention analysis, as long as the user had a return visit, it was defined as retention, which would lead to false high retention indicators.
05 Cluster analysis
Cluster analysis has simple and intuitive features. Clustering in website analysis is mainly divided into: user, page or content, and source.
User clustering is mainly reflected in user grouping and user tagging method; page clustering is mainly similar and related page grouping method; source clustering mainly includes channels, keywords, etc.
For example: In page analysis, there are often bands? Parameters page. For example: information detail pages, product pages, etc., all belong to the same type of page. Simple analysis is likely to cause inaccurate problems such as bounce rate, exit rate and other indicators. Through cluster analysis, accurate data of similar pages can be obtained for analysis scenarios.
06 AB test
One of the main ideas of growth hackers is not to make a big and comprehensive thing, but to constantly make small and precise things that can be verified quickly. Quick verification, how to verify? The main method is the AB test.
For example: You found a loophole in the funnel conversion. Suppose it must be a commodity price problem that caused the loss. You saw the problem-the funnel, and you came up with an idea-change the pricing. But whether the idea is correct depends on the real user’s reaction, so using the AB test, some users still see the old price, and some users see the new price. If your idea is really effective, the new price should have a better conversion. If this is the case, the new price should be determined and optimized repeatedly.
07 Buried point analysis
Only by collecting enough basic data can the required analysis results be obtained through various analysis methods.
Through the analysis of user behavior, and subdivided into: browsing behavior, light interaction, heavy interaction, transaction behavior, for the browsing behavior and light interaction behavior of the click button and other events, because of its frequent use, simple data, the use of buried point technology The realization of self-service burying can improve the effectiveness of data analysis, and the required data can be extracted immediately, which greatly reduces the workload of technicians and requires the behavior of collecting more abundant information.
For example, heavy interaction (registering, inviting friends, etc.) and transaction events (adding a shopping cart, placing an order, etc.) are implemented through SDK batch embedding.
08 Source analysis
With the disappearance of traffic dividends, we attach great importance to the source of customers. How to effectively mark the source of users is of vital importance.
Traditional analysis tools, channel analysis only has a single dimension. It is necessary to deeply analyze the effects of different channels at different stages. SEM paid search and other source channels and user regions are cross-analyzed to obtain detailed customer acquisition information in different regions. The more detailed the dimension, the analysis result The more valuable it is.
09 User Analysis
User analysis is the core of Internet operations. Commonly used analysis methods include: active analysis, retention analysis, user grouping, user portraits, and detailed user checks.
User activity can be subdivided into active browsing, active interaction, active trading, etc. Through the subdivision of active behavior, key behavior indicators can be grasped; grouping by user behavior event sequence, user attributes, and observing grouped users’ visits, browsing, and registration, Interaction, transaction and other behaviors, so as to truly grasp the characteristics of different user types, and provide targeted products and services.
The user portrait is based on the automatic labeling system, which clearly depicts the user's complete portrait, which more powerfully supports operational decision-making.
10 Form analysis
Filling in forms is an essential part of interaction between each platform and users. Excellent form design plays an important role in improving conversion rate.
From the moment the user enters the form page, there is a micro-funnel. From entering the total number of people to the number of people who finally completed and successfully submitted the form, how many people started to fill in the form during this process, and what difficulties did they encounter when filling in the form? Completing the form will affect the final conversion effect.
The above are common data analysis methods, and more application methods need to be applied flexibly according to business scenarios.