For Internet workers, data analysis is an essential skill! After spending so much money on , has the marketing effect been achieved? What kind of functions can really hit the pain points of users?
To learn data analysis, you must not only be able to use tools such as Excel, Python, etc., but also have a data analysis mindset. has compiled 7 of the best and most common big data analysis models for you! Help you complete data analysis efficiently!
7 common big data analysis models:
- event analysis
- retention analysis
- funnel analysis
- path analysis
- Session analysis
- distribution analysis
- attribution analysis
01 event analysis
What does do: studies the value of a certain behavioral event to the enterprise organization The impact and extent of the impact. How to use
: tracks or records user behavior or business processes, such as user registration, browsing product details page, successful investment, cash withdrawal , etc., by studying all factors associated with the event to discover the reasons behind user behavior events, interactive effects, etc.
application scenario example:
Question: The operator found that the PV data of a certain channel on a certain day was abnormally high. The reason needs to be investigated? It is divided into the following steps:
- Define events: Define events as PV values, and limit channels through filter conditions
- Multi-dimensional drill-down analysis: Analyze PV values according to different dimensions such as date, geographical location, machine model, operating system, IP, etc.
- Display the results in multiple dimensions and give an explanation of the PV data. Is it false traffic? (All from a certain IP), the value is abnormally high? (The data increased one day)
02 What the retention analysis model
does: analyzes user participation/activity and examines how many users who performed the initial behavior will perform subsequent behavior. This is an important method used to measure the value of a product to users. How to use
: There are three types of retention analysis, N-day retention, unbounded retention, and bracket retention.
Application scenario example:
Question: has a lot of new customers recently. I want to see how the user retention is? It can be divided into the following steps:
- Group the same period according to the time of new user registration/download (month/week/day)
- Observe the 7-day retention, 14-day retention or 30-day retention of the user's investment (you can choose freely)
- Compare different cohorts and observe the daily changing trends of retention rate and .
03 What does the funnel analysis model
do: reflects the user behavior status and the user conversion rate at each stage from the starting point to the end point. How to use
: compares the difference in the conversion rate of each process step by observing the conversion rate of each link of the user group with different attributes. , understand the user groups with the highest conversion rates, analyze the rationality of the funnel, and make adjustments for abnormal conversion rate links.
Typical scenario examples:
Question: sales staff want to know the conversion status of each link from demand discovery to winning orders? You can refer to the following steps:
- Start from the demand discovery process, design sales funnel diagram
- according to the demand, follow up and enter data according to each link of the sales funnel diagram
- automatically produce data analysis tables, observe customer conversion rates at each stage
Detailed reference:
04 What does the path analysis model
do: , a type of data analysis method unique to Internet products, improves the arrival rate of product core modules, extracts the mainstream paths of specific user groups and characterizes user browsing characteristics. How to use
: analyzes the user's flow patterns and characteristics of various modules in the APP or website based on each user's behavioral events in the APP or website, and mines the user's access or browsing patterns to achieve some specific business purposes.
path analysis is mainly to solve the following questions:
- Which path does the user mainly take to finally form a payment conversion?
- What is the actual direction of the user after leaving the expected path?
- What are the differences between user behavior paths with different characteristics?
05 session analysis What the model
does: performs indicator analysis on a collection of user behaviors that occurred on the website/H5/mini program/APP within a specified period of time. The key point of
Session is: what did the user do for how long. How to use
: includes the number of visits, the number of visits per capita, the total visit duration, the duration of a single visit, the depth of a single visit, the number of bounces, the bounce rate, the number of exits, the exit rate, the duration of visits per capita, the total page stay time, and the average page Length of stay, etc.
Two formulas:
- Average access duration = sum of session markets of all users/number of sessions
- Average interaction depth = sum of events within the session/number of sessions
06 Distribution analysis model
What does it do: distributes status through changes in quality Problems are found in the analysis to understand whether the production process is normal and whether scrap products have occurred. The tool is histogram , so it is also called the histogram method. How to use
: provides "dimension indexation" and then decomposes the data, divides the original dimensions into dimensions according to certain numerical intervals, and then analyzes the distribution of each dimension interval.
gives an example:
separates out the users who are particularly dependent on it, and builds a special user-operated project to operate the users. You can also use users with large payment amounts to do some operational activities.
The following figure shows the distribution of ’s number of people and transaction orders:
07 What does the attribution analysis model
do: The problem to be solved by attribution analysis is the generation of advertising effect , and how should its credit be reasonable Which channels is assigned to. How to use
: measures and evaluates the contribution of user touchpoints in the site to the achievement of the overall conversion goal, and can very directly quantify the conversion effect and value contribution of each operation position and touchpoint.
For example:
Xiao Chen saw an advertisement on Moments on his mobile phone, which released the latest Apple mobile phone. During his lunch break, he was browsing , Douyin, and saw an internet celebrity commenting on it. When he was browsing Moments on the subway after work, I found that some friends had already received their mobile phones and were posting pictures, so I drank a cup of Jiangxiaobai to strengthen my courage and went home to apply for funds with my wife. Finally, my wife approved and let him go to Jingdong to buy it, which was guaranteed. So, how much value did the various channels of Moments advertising, Douyin, Friends Moments, and JD.com contribute to this transaction?
I hope the above data model can help you get twice the result with half the effort! Follow @Jiandaoyun, a lot of useful information is coming~
