Analysis: how to do a good product manager, good data analysis?

(Research papers Download News) product manager in order to do data analysis, should have a complete system of thought, in terms of value, methodologies and tools three levels of reserve-related knowledge. At the same time based on the product and the user data to polish products, with data to test iteration, constantly improve the user experience.


This year, with the spread of big data, lean operations, growth hackers concepts, data analysis thinking is gaining in popularity. At the forefront of the Internet product manager who contacted a large number of user data, but it has been troubled by how to do data analysis.

Then the product manager how to build their own data analysis knowledge? Value data analysis, then where? Product manager for data analysis what specific methods? How learning data analysis? This article will share with you these issues.

Data analysis system: Roads, surgery, devices

'Road' refers to values. To product manager is to do data analysis, we must first agree with the meaning and value of the data. A do not agree with the data analysis, the significance of the lack of understanding of data analysis are difficult to do the job.

'Skill' refers to the correct methodology. Now emerging 'Growth Hacker' (Growth hackers) concept from AARRR frame (acquisition, activation, retention, liquidity and recommended five links) start with product analysis, this is a very good analysis.

'Device' refers to data analysis tools. A good data analysis tools should help the product manager for data acquisition, data analysis, data visualization, saving time and effort product manager, product manager to help users better understand and better optimize the product.

Value data analysis

Product manager for data analysis and analysis can not be, but the end result you want to put on the product and the user. Data analysis should help to optimize product design, product manager and iterative, drive products and subscriber growth.


When we line up on a new product (product) or function, it needs to be monitored and measured data (measure). Then collected from monitoring user behavior data products (data), and these data were analyzed and summarized (learn). Finally, the conclusions drawn from the analysis and opinion (idea), if the data proves that our new products / features are excellent, we can promote; data on our products if there are problems, we need a new round of product optimization (build).

In the 'product - data - conclusions' continuous cycle, we continue to use the data to optimize our products and accelerate the pace of iterations to enhance the user experience.

The method of data analysis

Method 1: traffic analysis

Analysis of different eligible quantity and quality of passenger flow channels, thereby optimizing delivery channels. Common approaches have UTM tags to track, analyze new user's ad source, advertising content, advertising media, advertising items, ad name and keywords.


Access real-time monitoring of movements of products, in particular care flow outliers. For example, an Internet platform as a financial product BUG cause the user traffic spikes caused by panic buying, product manager for real-time data found quickly off the assembly line after the abnormal product fixes BUG, to avoid further loss.

Method 2: Conversion Analysis

All commercial websites are broadly electricity supplier website; because users need conversion, the user needs to be realized. In our products there are many places to do the conversion Analysis: Registration conversion, later transformed to activate the conversion and the like. Usually we use a funnel to measure the user's transformation process.


Many factors affect the conversion rate, we summarize the three major aspects: the flow channel, the user marketing, website / APP experience. In channel flow, for example, to quantify distribution and delivery of our resources through preferred channels, can effectively improve the overall conversion rate.

Method 3: Analysis retention

Retained, by definition refers to the number of days after returning users first visit to your website / APP after. Retained the core product growth, the user only to stay, your product can continue to grow. A retention curve, if you do not do what product manager, then Movies tiger slowly lost.


From product design perspective, the key to finding the trigger behavior retained to help users find the product key nodes retained as soon as possible. Before we find ourselves inside the product, use the 'New' retain a very high degree of functional user; so we do a product improvement, the 'New' button in the top of the homepage to stimulate users to use, the effect is very good.

Silicon Valley, the popular Magic Number (magic numbers) are also retained part of the analysis, such as Facebook found that 'in the first week of Riga 10 friends,' the new user retention is very high. As a product manager, and we need to continue to explore our data analysis product inside the magic number, continuously improve customer retention and degree of activity.

Method 4: Visualization Analysis

The user experience is a very abstract concept, we can visualize them. A common approach is to present the user's data visualization in the form of heat maps presented.


With heat maps, product manager can be very intuitive understanding of the user clicks on the product preferences, test our product design or layout is reasonable.

Method 5: Cohort Analysis

Thousand thousand faces, product manager for user fine analysis is essential. Different regions, different sources, different platforms and even different types of mobile phone users are likely to exist a huge difference to their perception and use of the product. Product Manager users can be grouped different attributes, different groups of users to observe differences in behavior, to optimize products.


Before we conducted an analysis of the overall conversion rate of site was registered 6%; but the use of Chrome's new user registration conversion rate of 12%, using the IE browser's new user registration conversion rate was 1%. Such a point, the problem is very obvious, very likely a browser compatibility issues, product manager should look at this issue.

Data Analysis Books

Do data analysis, once a day you can not, you need to design in product planning, product upgrades iteration constant practice. Below these books, product manager for data analysis study has some help:

Recommendation 1: Fan ice 'growth hacker'

This is the first hacker to grow this detailed description, from the author's perspective AARRR cut, we described a number of product optimization, product growth case, very useful for product manager.

Recommendation 2: Eric Ries 'lean Data Analysis'

In this book, the author relates introduce relevant indicators data analysis, data analysis points in different industries, and a large number of cases of data analysis and informative data. If you want to analyze the data floor, the product manager for this book is very helpful.

Recommendation 3: Products and analysts GrowingIO write 'Internet growth in the first of the Data Analysis Handbook'

There is a compilation of more than a year of our data analysis, product optimization of actual cases, there are large number of articles have been turned, for example, 'how to be a good product manager data' and so on.

Download the electronic version of the manual analysis, please refer to here.

Recommended 4: Eric Ries of 'Lean Startup'

The authors suggest the minimum feasible product (MVP), small run, fast iteration and optimization of product design concept, far-reaching.


Data analysis is more than one discipline, cross knowledge in many fields, something very much involved. Product manager in order to do data analysis, should have a complete system of thought, in terms of value, methodologies and tools three levels of reserve-related knowledge. At the same time based on the product and the user data to polish products, with data to test iteration, constantly improve the user experience. (Source: GrowingIO compile: China Electronic Commerce Research Center)

[DBNETLIB][ConnectionOpen (Connect()).]SQL Server 不存在或拒绝访问。

Related Research Papers on data analysis

Internet Research Papers