Except for commercial data products, the UI resources of data products used internally by most Internet companies are very scarce. Because there are no better alternatives, internal users usually have a high tolerance for data products, and "usable" is the basic expectation of users.
A story about product requirements and development and implementation was circulated a long time ago (see the figure below for details). The core idea expressed is that the real development and implementation process will encounter various constraints such as time cost, resource cost, and development quality, and the final product will be consumer email list launched. There is a certain gap between the meeting and the product demand. To sum up, "the beauty of the product will not exceed the beauty of the product manager". For data products without UI/UE, the quality of the product PRD is the final upper limit of the product. If the product plan is very consumer email list rough, don’t expect too much to achieve the effect. So how to quickly design a high-fidelity data visualization product prototype?
Step 1: Identify needs
The basic requirement of the business is to be able to quickly obtain business KPI data and perform data analysis. Data visualization is to provide a more efficient solution for data readability, visualization, and interactive analysis.
For product managers or operations, what are their KPIs, what actions are used in daily work to achieve them, what data is needed to support decision-making/review in the process, and what are the common dimensions of indicator splitting? Taking product managers as an example, daily attention should be paid to different platforms, different traffic entry flows and conversion rates, the conversion funnel of the key path, and the decision support for product function iteration based on data.
Step 2: Determine the analysis idea
Visual products without analytical ideas have no soul, and the value of the product cannot be reflected only by stacking icons. A good visual product must have clear analysis ideas included in the product interaction design process. Combined with the example of traffic analysis, the product analysis scenarios are as follows:
Turn on the computer after going to work every day, and the product manager can most intuitively see the latest data of the market KPI (visit conversion rate), as well as the changing trend
Split it into different platforms such as App, WeChat applet, H5, etc. to see if the indicators are normal, and if there was any problem with the release of WeChat applet yesterday
It is found that the conversion rate of WeChat mini-programs has dropped by 20% from last week, and it is necessary to further analyze which traffic entry conversion rate has decreased.
After analysis, it is found that the conversion rate of the newly revised information flow mode at the bottom of the homepage has decreased by 40%, and then the conversion funnel of the whole process from access to ordering is further divided to see which step has the largest user loss.
To sum up, when designing a product, users should look at the data, analyze the data, and form a product interaction path, which can guide users to operate step by step, and finally locate the problem, instead of downloading the data, excel is doing screening and analysis.
Step 3: Design a product plan
1. Page Design
Common visual page designs include interactive analysis of indexed management and waterfall display of charts, each with its own advantages and disadvantages.
Layout mode 1: indexed management mode
The page is concise and the key points are highlighted. Obtain more data through dimension selection and tab switching to avoid too many charts on a page that are difficult to focus.
The analysis ideas are clear, from summarization to dimension subdivision, and split layer by layer when necessary
It is convenient for authority management and control, and index-based management can control authority from dimensions and indicators
The amount of information displayed by default is limited, the analysis process relies on interactive selection, and the information is hidden deeply
Index-based management is suitable for cases where the number of core KPIs is small (within 10), and the index dimensions should be unified. When different index dimensions are different, the interaction needs to be adjusted to a certain extent, that is, the date is used as a common dimension, and other filter conditions can only be adjusted according to the index tab. Toggles are located below the Metrics card.