Forum Posts

sifatahmed
Jan 13, 2022
In General Discussions
One of the most intriguing questions that product managers and marketers face is, ‘What are users doing on my app?’ Even tougher is how to extract actionable strategies from user data. Luckily, modern data visualization and analytics best practices exist to help better understand your users. Two potential methods for visualizing data are the sankey chart and the sunburst chart. Though the underlying data may be available, it becomes a tedious job to sift through tons of user data. Let’s take the simplest form of user journey in tabular data: The above table contains the first 5 sequence of steps users take after launching an app. It is very difficult to draw insights from such data. This is where visualization helps. With the use of the right visualization, we can summarize and effectively draw analytical insights from the data. In simple terms, delivering more insights per pixel is the goal of good visualization. What is a Sankey Diagram? A Sankey diagram is a type of chart that displays flows and their quantities. Arrows of various thickness are used to visualize the quantity in each flow as well as the direction or path in which they flow. You might typically see this diagram used to show the flow of money, materials, information, or energy. What is a Sunburst Diagram? A sunburst diagram, on the other hand, is a type of data visualization that is radial in shape. It Colombia Phone Numbers List is also known as a multi-level pie chart or a radial treemap. Each ring shows hierarchy with the center being the root. And rings can be sliced for each category. Data Visualization Constraints We faced 2 main constraints while selecting the right type of visualization: Capture maximum sequential data Since we are analyzing user journey across sessions, we will definitely be analyzing more than 10 sequences of events users perform. Hence, the visualization should be able to capture the maximum number of sequences possible. Space constraint The space required to depict the visualization should not be more than 70%-80% of user’s visible area of the screen. Given the above 2 constraints, we had to choose between two main type of plots: 1. Rectangular Plot – i.e. Sankey Chart 2. Radial Plot – i.e. Sunburst Chart The easier way to find out the winner was to apply the plots to actual data. As we experimented with actual data, we were inching towards Radial plot. Let’s take a couple of examples to highlight the benefits of the radial plot. Rectangular vs Radial Data Visualization Clock Example The above graphs show the heatmap of Videos played by Day of the week and Hour of the day. The radial heat map conveys the message where the heat is more concentrated much faster than the rectangular heat map. It is clear that the heat is relatively more concentrated between 8pm-10pm between Mondays-Saturdays whereas it is relatively more concentrated between 11am-7pm on Sundays. Key Learnings In both the examples, we had to deal with a lot of information or categories. The Clock had the first 12 integers depicting the hour, minute and seconds information. The heatmap had a total of 168 grids (24 hours * 7 days) to depict the information. In both cases, Radial plots were able to relay the required information faster. Now that we established Radial plots scored over Rectangular plots in the above use cases, let’s apply Sankey and Sunburst charts to the main problem statement of depicting user flows. Sankey Charts The current industry practice to visualize user paths/flows is a Sankey Chart, which is a rectangular plot. Sankey Chart In the above diagram, there is a flow of one set of values to another. The values connected are nodes and the connections are links. The width of the link signifies the relative strength of a flow from a node to another node. It gives a bird’s eye view of the journey and allows the viewer to interact and see the relation between certain nodes in the visualization. Plot Twist Takeaways from Chart Examples Dominant Nodes and Links are easily visible Easy to analyze sequences up to 4 levels. Dominant contributions to each node from previous nodes can be easily analyzed.
0
0
1
 

sifatahmed

More actions