What data capabilities do data product managers need to master?

With the continuous enrichment of big data technology and application scenarios , The value of data is valued by more and more enterprises , Even data driven 、 Data empowerment as a new growth point . At the national level, the data has also been upgraded to an important strategic asset , Data has become an important part of the new infrastructure . Then comes the data product manager , Gradually become the digital transformation of enterprises 、 A necessary position in the process of data operation . In past articles , Capability model for data products , As well as the classification of posts, I have done special science popularization , As the name suggests , And others C End 、B The biggest difference between product managers on the end is the processing of data raw materials or processing tools , So I want to make another introduction to the data capabilities that I need to master , For new people who want to work as a data product manager , Provide some suggestions on the direction of preparation .

One 、 Data acquisition and embedding point

The essence of data-driven is data-driven growth , In the past, more attention was paid to the order volume 、 Macro performance indicators such as revenue , Data operation from macro to micro , Start to pay attention to the whole process behavior insight of users from recognizing products to finally completing a business process , And in the process , Embedding point is the starting point of data value , Data products are based on data , Do user behavior analysis or visual analysis of data products , As data product manager , We need to standardize the management of buried point data , Otherwise, there is no buried point to collect data , It's hard to make bricks without straw , The designed product scheme is perfect , Interactive rewiring , There's no data . therefore , We must master how to promote business products and research and development, and choose appropriate and efficient methods of embedding points , And establish unified specifications and processes , Control the data quality of buried points .

1. Common buried point scheme comparison and type selection suggestions

See previous articles for details ： User behavior data collection ： Comparison of advantages and disadvantages of common buried point schemes and suggestions on type selection

2. Clear responsibilities , Establish an efficient buried point collaborative workflow

In agile work theory ,80% All the problems are process problems , It can be avoided and solved through standardized processes , In the past, I experienced many pits about burying points , The right and responsibility are not clear . Although the implementer of the embedding point is business research and development , The users of data are business products and operations , But it is the data products and data departments that are ultimately responsible for producing data , Although you can say “ Business is not buried , I can do nothing about it, either ” Throw the pot out , But if no data is available , What about drive empowerment , Then the data team can also be disbanded . In a word , Data products should shoulder the responsibility of formulating buried point specifications , Cooperate with all parties to establish standards , Finally, the standard can be integrated into the buried point management system , Process 、 automation .

Two 、 Data analysis ability and index system construction

Data is like crude oil , Few people can use it directly . Only through analysis 、 Hand of excavation , Purified into gasoline 、 diesel oil 、 And all kinds of chemical fiber products , To maximize value . After collecting data at the burial point , Analysis is also needed 、 abstract 、 Commercialization , Can be directly used by more businesses . In the process , It needs to have an indicator system for building business monitoring , And the ability to integrate analytical ideas into data products .

1. Ideas and methods of index system construction

In fact, various data official account 、 We media has a lot of methodology for index system construction , Different paths lead to the same goal, and there are little differences . Here is a brief overview and summary .

（1） What is the index system ？

Based on the needs of business operation and development , Will be able to comprehensively measure the indicators of business health （ Index library ） According to a certain relationship （ Connections ） Combined to form a systematic index management system , And each index can judge whether it is good , Healthy or not （ Evaluation criteria ）, Once problems are found, they can be broken down through indicators or dimensions （ Analysis dimension ） Focus on the problem .

（2） Why ？

A single indicator cannot be accurate 、 Comprehensively measure business health , And lack of related indicators , It is difficult to focus when there are too many , Lost focus . If there is no evaluation standard , Then the indicator is just a data , It cannot be called knowledge or information . It's hard to really drive decisions .

（3） How to construct the index system

General principles ： Starting from the key objectives of the business development stage , be based on OSM Model 、UJM( User life journey )、AARRR（ Pirate model ） Wait for the model , Index and monitor the business process , And classify according to the split relationship of indicators .

3、 ... and 、 Integration of data products and analysis ideas

Many new data product managers are doing data visualization products , It is easy to pursue visual effects , But it ignores the soul of visual products , The value of data products lies in mining the value of data , Let more non data professionals 、 People without data awareness can quickly get decision-making information from data products , Not just as a data retrieval tool . Judge a data visualization product or Dashboard Usefulness 、 The standard of eligibility is to do ： What is the data , How about the data , What's the problem , How to solve the problem . Common analysis methods, such as ： comparative analysis 、 Funnel analysis 、 Trend analysis 、 Composition analysis, etc . How can data visualization have a soul

Four 、 Theoretical basis of data warehouse

Business data 、 Data capitalization , Hierarchical management of data warehouse can not only improve the reusability of data , It can also improve the efficiency of data operation and maintenance . When making data asset management and governance tool data products , Integrate the standardized process of asset reduction into the system , Every time you create a table , System based prompt and automatic filling can be fast and efficient , Instead of requiring every developer to form rules into muscle memory . Besides , For data consistency 、 integrity 、 timeliness 、 Accuracy and other data quality monitoring rules also need to be mastered , After all, data quality is the lifeline of data products , If the data given to the business is wrong , Decision making mistakes 、 Precision marketing is no longer accurate .

5、 ... and 、 Big data commonly used technology and data flow link

Although I say “ Everyone is a product manager ”, In the past, there were many disputes about whether product managers should understand technology , Personally, as a data product manager , At least understand some basic big data technology application scenarios , And the link of data flow , Such monitoring of data quality , real time 、 Differences and technical limitations of offline data , Design products more reasonably . And for the underlying development suite 、 The data product manager of development tools is even more necessary . Just imagine , If you even HDFS I don't know what it is , When you need to design a HDFS Tools for directory management , Where should you start .

6、 ... and 、 summary

The particularity of the position of data product manager , It is doomed to have a hard relationship with data . To be an excellent data product manager , In addition to having the general ability of product manager , We should also establish our core competitiveness in the data direction , Only when doing data products like this , Will be more handy .