Lei Feng Network: Author Chen Ming, GrowingIO co-founder & Vice President of Operations, graduated from Stanford University, has worked in eBay, LinkedIn Data Analysis Department, has a wealth of business analysis experience.
In recent years, there has been an increasing demand for data analyst positions from Internet companies. This is no accident.
Over the past decade or more, China’s Internet industry has grown wildly with demographic dividends and flow dividends. With the rising cost of traffic acquisition and declining operational efficiency, this extensive business model is no longer feasible. Internet companies urgently need to use data analysis to achieve refined operations, reduce costs, and increase efficiency; this places higher demands on data analysts. This article will share with you the evolution of data analysts, data analysis value system, the data analyst's four major capabilities, seven commonly used ideas and practical analysis cases.
I. Past and Present of Data AnalystsBefore introducing data analysts, let's look at these historical figures first and see how they all relate to data analysts.
The famous "analyst" in history
The six historical figures shown above (from left to right and from top to bottom) are: Zhang Liang, Guan Zhong, Xiao He, Sun Bin, Gui Guzi and Zhuge Liang. They are the most famous advisors in history, and some have also done prime ministers. They have read a lot of books and have a unique perspective. They have discovered many laws by summarizing a large number of historical facts and successfully predicted many events in practice. They have created tremendous value for their organizations through the practice of “history statistics—summarizing analysis—predicting the futureâ€. This is the predecessor of “data analystâ€.
So now, what are the necessary skills for data analysts and how can they become a good data analyst?
Second, the data analyst's value pyramid A complete enterprise data analysis system involves multiple stages: collection, cleanup, conversion, storage, visualization, analysis and decision making. Among them, the work content of different links is not the same, and the time consumed and the value generated are also far from one another.
Data analyst's value pyramid
There are at least three aspects of data in the Internet enterprise data analysis system: user behavior data, transaction order data, and CRM data. Engineers collected data from different sources and then unified them to the data platform through cleanup, conversion, and other links; then specialized data engineers presented data from the data platform. These jobs take up 90% of the entire process, but only 10% of the value is generated.
The data analysis of this pyramid is further closely integrated with the actual business. It supports the business decision-making of the company through reports, visualizations, and other methods, covering all frontline departments of products, operations, marketing, sales, and customer support. This part takes up only 10% of the entire link, but it can generate 90% of the value.
A good business data analyst should be value-oriented and closely integrate product, operations, sales, and customer support practices to support business lines to identify problems, solve problems, and create more value.
Third, the data analyst's four major capabilities
Data analysts must have four capabilities
1. Overall viewOne day, the product manager ran over and asked me: Hi, can you help me see the data sent by the new product features yesterday? Thank you! Reflex I will say: Well, I'll give it to you right away! However, I still politely asked: Why do you need this data? The product manager replied: Oh, yesterday the new features went live and I wanted to see the effect. Knowing the purpose of the product manager, I can carry out data extraction and analysis on a targeted basis, and the results and recommendations of the analysis will be more maneuverable.
In many cases, data analysts cannot count on numbers and fall into various reports. A good data analyst should have a holistic view. He should step back and ask why when he analyzes the requirements, and understand the problem background and analysis goals better.
2. Professionalism
A company's data scientists are modeling and predicting user churn. The resulting user churn model has an accuracy of more than 90%. The accuracy rate is so high that business analysts can't believe it. After testing, it was found that the data scientist's model has an argument of "whether the user clicks the cancel button". Clicking on the "Cancel" button is an important symptom of user loss. The user who has done this action will almost always be lost. Use this independent variable to predict the loss without any business meaning and operability.
Data analysts want to demonstrate her/his professionalism in the industry (eg, e-commerce, O2O, social networking, media, SaaS, mutual funds, etc.), familiarize themselves with the meaning behind the business processes and data in their industry, and avoid the above data joke .
3. Imagination
The changes in the business environment are getting faster and more complex. The influence factors behind a set of business data are unimaginable to ordinary people. Data analysts should use their imaginations based on work experience to make bold innovations and assumptions.
According to Silicon Valley’s core KPI (Facebook’s 4-2-2 guideline, LinkedIn’s connection law), we also want to find the core KPI that drives the growth of Internet companies. Based on our imagination and the advantage of "billless" full-quantity data acquisition, we created "GrowingIO retained magician". Through data collected in full, intelligent back-end calculations, and simple interactions, retention wizards can help companies quickly find the user behavior most relevant to their retention, just as a magician can simply swipe a magic wand. For example, in a SaaS product, the retention rate of users (groups) who created 3 charts in a week is very high. Then "Weekly + 3 + charts" is the magic number that we drive user growth.
4. Confidence
Take the sales job as an example. A salesperson must first establish trust with the user; if the user does not trust you, then he can hardly trust or purchase your product. Similarly, data analysts must establish good personal relationships with colleagues in various departments to form a certain level of trust. The colleagues in each department trust you, and they may be more likely to accept your analysis conclusions and recommendations; otherwise, it will be less effective.
Fourth, the data analysis of the common seven ideas 1. Simple trendReal-time access to trends to understand product usage makes it easy to quickly iterate products. The three indicators of the number of users visited, the source of access, and the behavior of accessing users are of great significance for trend analysis.
Minutes of real-time trend analysis
Weekly trend comparison
2. Multidimensional decomposition
Data analysts can disaggregate indicators from multiple dimensions based on analysis needs. For example, browser type, operating system type, access source, advertising source, region, website/mobile application, device brand, APP version, and so on.
Multi-dimensional analysis of access user attributes
3. Conversion funnel
According to the known conversion path, the conversion of the total and each step is analyzed by means of a funnel model. Common conversion scenarios include registration conversion analysis and purchase conversion analysis.
Funnel analysis shows the churn rate at each step of registration
4. User grouping
In refinement analysis, it is often necessary to analyze and compare groups of users who have a specific behavior. Data analysts need to use multi-dimensional and multi-indicators as a grouping condition to selectively optimize products and improve user experience.
5. Check the path
Data analysts can observe the user's behavior trajectory, explore the user's interaction with the product, and then find problems, inspire inspiration, or test hypotheses.
Analyze the user's behavior through detailed investigation
6. Analysis of retention
Retention analysis is the exploration of the relationship between user behavior and return visits. The retention rate that we generally talk about refers to the percentage of “new users†who “return to websites/apps†over a period of time. The data analyst finds the growth point of the product by analyzing the differences in the retention of different user groups and using the differences in the retention of users with different functions.
Remaining analysis finds that users who create “charts†have higher retention
7.A/B test
A/B testing is parallel testing of multiple programs at the same time, but each program has only one variable different; then some rules (such as user experience, data indicators, etc.) wins and selects the optimal solution. Data analysts need to select reasonable grouping samples, monitoring data indicators, post-event data analysis, and different program assessments in the process.
V. Data Analysis Case: EDM Conversion AnalysisA social platform launches paid premium features and pushes them to target users in the form of EDM (Email Direct Marketing). Users can click on the link in the email to complete the registration. The registration conversion rate for this channel has been between 10% and 20%; however, the registration conversion rate has dropped sharply since late August, even to less than 5%.
If you are the company's data analyst, how would you analyze this problem? In other words, what factors may cause a sharp drop in EDM conversion rate?
A good data analyst should have a holistic view and professionalism, start from the actual business, and synthesize the possibilities in various aspects. Therefore, the possibility of a sudden drop in EDM registration conversion rate is as follows:
1. Technical reasons: ETL delay or failure, resulting in the lack of front-end registration data, a sharp drop in registration conversion rate;
2. External factors: Are there any holidays in this time node, and whether other departments have recently sent promotional e-mails to users, these factors may dilute the user's attention;
3. Internal factors: whether there is a change in the copy and design of the e-mail; whether the e-mail arrival rate, open rate, and click-through rate are normal; and whether the e-mail registration flow is smooth.
After one by one investigation, the data analyst locked the reason on the registration process: the product manager added the contents of the bound credit card in the registration process, resulting in a significant decrease in the user's willingness to submit the registration, and the conversion rate plummeted.
A seemingly simple conversion rate analysis problem, behind which is the manifestation of all aspects of data analysts. The first is the technical level, the understanding and understanding of ETL (data extraction-conversion-loading); it is actually a global view, and has a clear understanding of the business of seasonality, corporate level, etc.; finally is the degree of professionalism, and the flow of EDM business. , design and other well-known.
The power of practicing the raw data of data analysis is not a success, but it is constantly growing and sublimating in practice. A good data analyst should be value-oriented, look at the overall situation, based on business, and people are good, use data to drive growth.
Note: The original was published on the GrowingIO Technology Blog and the WeChat public number and was authorized to release Lei Feng Network (search for "Lei Feng Net" public number concerned) .
Video equipment cabinet,Floor standing equipment cabinet,Equipment and equipment cabinet,Multifunctional component cabinet
Jiangsu D-Bees Smart Home Co., Ltd. , https://www.cI-hometheater.com