1 Definition of data products
? Simply put, data products are products that can help users leverage the value of data to assist users in decision-making or actions, and use data as the main automated output. form. The emphasis on automated output is to distinguish consulting reports and manual reports from similar data research consulting companies. Next, data products can be refined:
? Can be divided into three categories according to user groups:
Internal data products of the enterprise, such as our company’s BI and automated approval mentioned at the beginning , data monitoring, etc.;
Commercial data products, such as Google’s Google Analytics;
Available to all users, such as Taobao Index.
2 The significance of data products
When we launch a new product function, whether it meets user expectations and whether it is popular with users, we need to speak through data . At Facebook, the growth team that reports directly to Zuckerberg has two data teams that collect, calculate and display data. They are responsible for the monitoring of all Facebook data and the ongoing optimization based on performance. An example of Facebook's emphasis on data-driven development is a good example. A team of 30 people led by a VP once spent a year revamping the homepage, but within three months of Grayscale's launch, it was rolled back due to poor data performance. Facebook uses quantifiable data to provide objective feedback on new features to drive next product decisions.
? Peter Drucker has a famous saying: if you can't measure it, you can't manage it.
3 How to design data products
? We can break this question down into five steps to answer:
-What users and scenarios are it for?
-What problems are solved or what value are brought
-What is the analysis idea
-What indicators are used
-How to combine and display these indicators
1) For what users and scenarios
? Product design must first clarify the users and scenarios for which everyone is familiar. The specific characteristics of data product users and scenarios are:
? - Different users have different values: especially for users Internal products of the company. From the perspective of the value that data can generate, a correct decision by the top management can save countless costs below. The value of a product cannot be measured simply by the number of users using the product;
? - Data that users at different levels care about Different granularity: When designing products, you need to always keep in mind the primary and secondary data presentation, analysis at different granularities, and the finest-grained entry. The essence of data analysis is to continuously segment and track changes;
- Different types of users use data in different scenarios, and the design should be centered around these scenarios. For example, our branch managers and team managers are busy at work and rarely work, so mobility and automation are critical. When designing, the principle is to display key indicators through the mobile interface, and the analysis results are brief and clear, with few analysis functions.
And when certain indicators change, you can be notified through your mobile phone in time. For data analysts in the office, the PC interface has more detailed analysis and comparison functions. The upcoming management decision-making system fully considers these scenarios. The business line browses and views through the mobile APP and supports notifications of indicator changes; another business data monitoring product presents detailed analysis on the PC interface. Only by fully understanding your users and usage scenarios and maintaining long-term and effective communication can you design better-use products.
2) What problem does it solve or what value does it bring?
That is, it is clear what needs the product needs to meet the user, and what urgency and value it has.
? First determine the user’s essential needs, which can be analyzed using the Demand/Want/Need method. The user says a cup of Coke (Demand). If what he needs most is to quench his thirst (Want), then a bottle of mineral water or Pocari would be more suitable for him (Need).
? Secondly, judging the value of the demand is based on two points: whether the demand satisfies core users; whether it is a rigid demand. The core user measurement formula is "number of people * single user value". We must have the concept of "not interfering with the normal use of core users for the needs of secondary users", let alone because some data products are only used by a few senior executives of the company. Lack of sense of achievement; rigid demand can be judged by comprehensive consideration of factors such as whether there are alternatives to the demand, frequency of occurrence (which can be considered based on when and where the situation occurs), duration, and other factors.
3) What is the analysis idea?
? What kind of idea should be used to analyze after the problem is clarified? The following principles need to be clarified:
? -Data product managers must have data analysis capabilities in order to better create greater data value;
? -Data product design concepts should start from overview to Segmentation and continuous comparison in multiple dimensions;
? - The design of the overview page of data products should be outlined, concise and clear in order to help users quickly locate and understand important information, data and important abnormal issues, rather than being immersed in them. Among the disorderly and complicated data details;
? -The subdivision of data should provide enough rich dimensions to facilitate analysis. Each segmentation must be carried out with indicators, and the results of all analysis must be implemented into action execution and closely related to the business;
? -The data itself is meaningless, and the comparison of the data is meaningful. The core of data products is to highlight contrast. This is the differentiated ability of data product managers, and the requirements are very high (as shown in the figure below for business data monitoring products). It requires both rich product design experience and deep business understanding and data analysis capabilities.
4) What kind of indicators are used?
? Analysis ideas need corresponding data support, and it is necessary to confirm that the data is accurate and complete, including what data indicators, data sources and fields are needed. During the confirmation process, you should pay attention to the following two points:
-The completeness of the data requires making it clear in advance whether all the data has been completely prepared. Data collection, cleaning and aggregation are the core contents of the data preparation process. If the required data is not collected in time or has not been cleaned, it will increase the risk of the entire project period.
- Many times when the accuracy of the data is used, it is discovered that the way of burying the points has always been wrong, or that the method of calculating the indicator does not exclude certain factors. The reasons are mostly due to the numerous and complex calibers of departments and the lack of unified data definition and quality monitoring and management. In order to solve this problem, the company is led by the data analysis center, and the product and business departments are involved in sorting it out. At the same time, in order to make the data collection more complete and accurate, we are also actively seeking changes in the selection of data collection points, and cooperate with Shence Data, an excellent third-party data service provider.
5) How to combine and display these indicators
? Regarding what kind of product form combination is used for data products to display indicators, common data product forms focus on data presentation, such as email reports, Visual report type, early warning prediction type, decision analysis type, etc.; focusing on algorithm type, such as user portraits, matching rules, etc.
Let’s discuss the product form design ideas that focus on data presentation:
? -Indicator design, first of all, it is necessary to clarify what type of products are suitable for which indicators, such as orders at the core of the project Conversion rate, loan amount, overdue rate, etc.
Split it layer by layer, no duplication or leakage. For example, overdue analysis can be broken down into overdue rate, number of overdue transactions, and overdue amount. Each node can also be subdivided into overdue distribution. Different products and different cities will have different overdue performance, and the overdue performance can be broken down layer by layer. ;
Ensure that indicators can clearly express their meaning and provide a basis for upper-level analysis ideas;
Clarify indicator definitions, statistical calibers and dimensions;
? -Indicators Presentation, data visualization. It is not only the job of UI designers, but also places high demands on data product managers. Because it involves how others understand and use your product. On the one hand, you need to continue reading relevant professional books, and on the other hand, you need to observe and learn enough excellent data products. Specific to the design of data visualization charts, some experiences include using curve charts for trends, stacked charts for proportion trends, bar charts for completion rates, bar charts for completion rate comparisons, and scatter charts with multiple indicators intersecting. Choose the appropriate form based on different indicators.
4. What are the requirements of data products for data product managers?
? We talked about data product design earlier, so how to ensure that the data products that meet the needs of the enterprise are designed, developed and launched smoothly according to the correct requirements? Woolen cloth? That is, what abilities do data product managers need to be competent? Generally speaking, there are four aspects of the ability model:
? -Data analysis ability: You must understand analysis, otherwise you will become a person who only produces reports. To transmit the microphone, you must understand the logic of data generation and be able to establish a data indicator system for the business module. Otherwise, the things that come out will be messy and may not be online for a long time;
? - Data display Ability, that is, the ability to visualize;
? - Ability to understand business models: Business theory must be understood before it can be abstracted into reports and analysis pages, and different business theories are suitable for different enterprises and different enterprises. In this stage, in addition to maintaining continuous learning and updating of business theories, it is also necessary to choose implementation based on the actual situation of the enterprise;
? - General product manager capabilities such as demand analysis and research, logical communication, rapid learning, etc.;
? In addition to having special requirements for capabilities, sorting out data indicators is a very tedious task during the design and development process of data products. In addition, when conducting data analysis, it is very likely that you will have to dig through a large amount of data. There is no result after half a day. This requires the data product manager to have a calm mind and be able to withstand loneliness and boring work. Therefore, the data product manager has some different personality requirements than the general product manager. For example, generally product managers are expected to be able to play games and have an extroverted and active personality, while data product managers are expected to be calm and reserved.
From the competency model of data product managers, we can see that they must understand data, products, business, and personality, which is quite demanding. The company launched its data strategy in early 2017. It can be seen that the formation of a data product manager team is a challenging task. In the hot data talent market, the formation work was completed in August last year. Comprehensive data product managers are hard to find, but we strive to form a comprehensive and combative team to make excellent products