Big data in this blue ocean, data product managers are on the head.
Editor’s note: This article is from WeChat public account “Everyone is a product manager” (ID: woshipm), author Anne.
Many people think that data product managers are product managers who deal with data, and they are not right.
The position of data product manager is actually cross-border: it needs to do data collection and cleaning; it needs to understand the internal and external user needs and understand the market; it needs to prove, falsify and discover problems by means of data.
——With so much work, what should the data product manager’s competency model look like?
In recent years, with the increasing attention of big data, artificial intelligence, and refined operations, major companies have become more and more popular with data processing and analysis applications.
According to the China Institute of Information and Communication Research, the size of China’s big data industry in 2018 is estimated to be 540.5 billion yuan, up 15% year-on-year; in 2019, it is expected to reach 621.6 billion yuan. In the next few years, it will maintain a growth rate of 10%-15%.
When the ecology of China’s big data industry continues to improve and the industry’s integration and application continue to deepen, it is undoubted that this will be a huge blue ocean.
This has also given rise to new career opportunities in the Internet profession – the professionalism of data product managers.
1. What is a data product?
To clarify the data product manager, the first and foremost question is “what is the data product?”
The author of the book Data Product Manager’s Practice Manual gives this definition:
Data products are a form of data value that can help users make better decisions; they can provide more in the user’s decision-making and actions. More analysis and data insights make data drive business more intuitively and efficiently.
From this perspective, search engines, personalized recommendations, Baidu Index, Taobao Data Cube and internal data support systems are all data products.
According to the flow of data, we can divide it into three levels:
1. Data Quality Products
To be a data product, you must first have data.
There are two ways to get data sources: others give them and find them.
-
Others give: such as advertising platform, user’s browsing data, access data and other behavior data in various applications; their own products are insufficient, obtained through API interface or other forms. But other people’s data quality usually has problems, garbled, missing fields is a common thing;
-
Looking for yourself: The data generated in your own system, such as the user’s travel trajectory of drip, the distribution of driver’s vehicles, etc., are all on their own platforms, which can be obtained by adding and burying points; There are various problems of data instability and inaccuracy.
The core of data quality products is to solve the problem of data reliability and stability.
2. Data Tools Products
Basic data already exists, but there are various problems with the data, and in order to reduce the calculation of the application layer, we usually complete various data conversion and services.
Simple understanding is: data cleaning according to different business needs, and then import data into various data conversion or calculation models, and provide data services for more downstream applications.
The model here is not necessarily a user portrait, a recommendation, but also a basic screening, sorting, matching, and simple logical calculation.
Summary in one sentence: The core goal of data tools is to improve the efficiency of data acquisition and make decision analysis faster.
3. Data Applications
The data applications of individual companies vary widely, and in general, they can be divided into two broad categories:
1) Analysis class
Analyzed products that help businesses analyze and make decisions through the calculation and presentation of data.
Typical: Traffic Analytics Platforms – Helping product managers with page design, functional improvements and revision assessments; and sales analysis products – can help operational analysis and increase sales conversion rates.
In addition to this, there are supply chain analysis systems, customer service analysis systems, and member analysis systems.
2) Algorithm class
The algorithm class directly changes the logic of the page through the calculation of the data.
TypicalThere are: personalized recommendations, programmatic buying ads, anti-cheat systems and other products.
Of course, these two categories are not strictly differentiated—continuous evolution, integration support.
In short, the core of the data application product is the data product solution for deep integration scenario applications, which is used to drive business development or realize data realization.
▲ Mature company’s data product architecture
As far as the above three types of products are concerned, data quality products and data tool products are more to ensure the stable, safe and efficient access of data. This is also the focus of the companies in previous years. It can be said that it is a data product. Underlying construction;
Data application products are mostly products that are known for their personalized recommendations such as today’s headlines, quick hands, etc. Currently, they are the core links of each key investment, how to better integrate the scene application and realize the realization of data value.
Second, what problems can data products solve?
From data quality products to data tool products to data application products, each level looks huge, so what value can it bring to our products and business?
See a few specific scenarios:
Products face these problems every day: How does traffic surge (or plummet)? What is the effect of the new channel? How does the user’s ARPU or per capita PV rise (lower)?
This is the problem solved by the data tool product-self-service analysis platform, which designes various indicators of products or functions (including income, DAU, ROI, etc.) into fixed reports and refines them into multiple dimensions (eg time, Areas, channels, etc.).
Based on this data, you can discover the anomalies in the data in real time, and quickly trace down to find the real reason for rapid adjustment to ensure that products and functions are on the right track.
Make business process data visible and help make business decisions scientific – this is the most basic use case for data products.
The headlines and quick hands that are known for their algorithms are not unfamiliar to everyone – when you click on an article/video of interest, the system automatically recommends more content that matches your interests. As time goes on, the accuracy of the recommendation is getting higher and higher.
How long does it take 200 million days to live, and the scale of daily average 10 million video uploads, if not based on big data recommendation system, how to rely on human resources?
The same is typical of big data-based intelligence.Marketing, intelligent scheduling, etc., are already typical applications of data products.
The high level of integration of data and algorithms, intelligent operation, and improved operational efficiency are the primary performance of big data.
Look at the “weather forecast” typical data product, using a long period of temperature, humidity, wind, daylight intensity, UV intensity, PM2.5 value, location information, various collections on the satellite Data, as well as meteorological data for various professions (examples, professionals please add it yourself); a series of “processing processes” such as screening, cleaning, analysis, and mining of these data can be obtained in several core meteorological features in the next few days. Data values and probabilities (temperature, wind, rain, snow, etc.).
The weather forecast we see is a combination of the above core information, giving video + GIS display, and giving users advice on “action” (travel advice, dressing index, Car wash index, etc.).
In addition to the above cases, there are bus route planning and bicycle lane design based on traffic travel data.
The higher value of data products, it is no exaggeration to say that it is to derive more data applications and realize the realization of data value.
So in our opinion, the data product is also equal to “application scenario + data + product”, and the data is empty talk about it from the application scenario.
Third, the data product manager’s ability requirements
Return to a specific business scenario, what are the data product managers doing, and what difficulties and challenges will they encounter if they want to be a good data product manager?
1. Data Product Manager’s Work Content
We interviewed data product managers from a number of companies and found that they are data product managers, but they are not the same in different companies.
Now, I understand the current data product manager from both a narrow and broad perspective.
In a narrow sense, a data product manager is responsible for implementing a data product tool and using it to meet specific data usage needs; that is, to undertake the data quality products, data tool products, and data applications mentioned in the first section. Product planning and design work.
In a broad sense, data product managers are not limited to implementing data product tools, but also data-related work such as data analysis and operations, and responsible for the company’s data services.
There are two main categories of work:
-
Data production: write some data production scripts, output data sheets, and even maintain data production processes;
-
Data extraction: responsible for extracting data from data requests from the business and delivering accurate and reliable data;
-
Data Analysis Report: Analyze daily business, output analysis reports, and form business conclusions;
-
Data Operations: Building indicator dictionaries, operational indicator dictionaries and data products, operational data, troubleshooting data issues, etc.
2. Data Product Manager’s Competency Requirements
Based on the recruitment requirements and job responsibilities of major manufacturers, the clustering of each skill tag is integrated. We find that the data product manager’s ability model, in addition to the basic skills of common product managers, data capabilities, business abstract understanding and project coordination Management capabilities are especially important.
1) Requirements for data capabilities
For data product managers, data capabilities are not a cool form on R Studio, nor a few PivotTables in Excel, and certainly not a few SQL to extract a few numbers; A set of analytical methods from the company’s business competition strategy to first-line business operations.
Based on this approach, data product managers can analyze the company’s different top-down businesses into the same model to help decision-makers quickly locate problems through data.
This requires data products to have very strong data sensitivity and data thinking – including index dictionary design, buried point design, data production related knowledge, data analysis and so on.
Of course, the basic tooling skills package is also a must-have – such as SQL, Excel, Python and other common tools for data processing.
2) Requirements for business abstract understanding capabilities
Business abstraction comprehension, the ultimate goal of data product managers is to make the data self-expression, to provide business-based log-based, from reporting to intelligent forecasting tool suite.
On the one hand, products must continuously extract and integrate all kinds of data from various businesses; on the other hand, the data should be expressed stably and quickly through various tools, so that the business can easily and quickly obtain insight from the data.
In order to do this, product managers must have the ability to understand business abstractions: know how to abstract difficult data from relatively closed business, open it to the business in the form of services, and ultimately form data supply to data. The closed loop of the application.
3) Requirements for soft skills such as project coordination management
As a product that consumes a lot of development resources, good project management capabilities determine the success or failure of the entire data project.
Data products face a very complex collection of data streams and business flows. In order to make a report system with complete data, stable system and fast query, the product manager needs to be from the bottom.The log starts to sort out and sort through the various data processing processes.
Because of the logic of processing, the development of data systems usually takes a long time period, and the demand for data in the business is usually either absent or unreasonable.
In this process, the coordination of data product managers and engineers, business parties is very challenging – both the logical abstraction and the hard power of data thinking, but also the soft skills of communication and collaboration, which may be data The most challenging place for product managers.
4. Conclusion
For data product managers, the content mentioned above is only to help data product managers, can have a framework understanding of the challenges and work of the data product position.
However, there are more problems that need to be solved in the actual business scenario, which is limited by the length, and will not be expanded here.
For data product managers, the competency requirements are very large:
-
On the one hand, it requires strong data analysis capabilities and logical abstraction. It can understand the various process steps in the business, track the flow of data, and understand the direction of business needs, thus transforming into product functions for business. Used;
-
On the other hand, it requires strong communication and coordination capabilities and project management and control capabilities, and can promote the implementation and use of related functions in the context of multiple lines of business, different needs and backgrounds.
Is there a challenge, but why not?
With the gradual development of 5G and the Internet of Things, we are experiencing a singularity in the development of science and technology. It is foreseeable that the application and value brought by data will far exceed the previous imagination. Maybe you are now The ceiling in the future is a new step, and the development of data product managers will far exceed your expectations.
The last reference to the father of high-tech marketing magic, a question from Silicon Valley strategy and innovation consultant Jeffrey Moore:
Without big data, you are blind and deaf and in the middle of a freeway.
Thanks:
“Data Product Manager Practice Manual: From Zero-Based to Big Data Product Practice” by @梁旭鹏
Drip / Suning / Jingdong / Taobao and other product managers who participated in the interview
Know the column “Data Product Manager’s Job Type and Competency Requirements” by @Blazeer俊辉
https://zhuanlan.zhihu.c