The wrong way to build a dashboard
An essential element of Tableau’s value is delivered through dashboards. Allowing the audience to interact with a dashboard and change the details being displayed provides a means to shift context—leading to new and potentially important discoveries.
A dashboard is only effective if it answers useful questions for people. That’s why a customized, user-specific dashboard is more impactful than a ready-made, one-size-fit-all report.
Traditional ways of providing reporting tools have been done by companies that have a core competency in data collection and storage. These entities attract people that are very knowledgeable in the technical aspects of database like building data quality, and data storage, but not data presentation.
Traditional buyers of business information systems tend to be people from finance, accounting and the information technology group because they possess the technical knowledge of database design, data collection, and data governance. Plus, IT is generally responsible for installing and maintaining the system.
Neither of the group possesses knowledge of the best practices related to data visualization. Their knowledge of charting typically comes from the commonly available spreadsheet programs, which often provide a lot of unnecessary and inappropriate chart styles. Historically, older business information (BI) tools that information technology staffs are familiar with, for report building have been more adept at data creation and storage but not information visualization.
Good report builders from both of these groups develop time-saving techniques that work well for creating dashboards in old-style tools. Unfortunately, those techniques are more concerned with the technical challenges of building the report and not on the aesthetic qualities of the user experience.
Why would an experienced designer use overly complex graphics? One reason this happens is because dashboards created with legacy tools are more difficult to build as they require more time and effort to produce. Often with legacy tools it makes sense to place as much information as possible into a single view to save time. This practice can lead to visualizations that are complex and difficult for end users to understand. Also, internal customers ask for what is familiar (grids of numbers) so that they can receive what they understand. Unfortunately, these techniques are exactly the wrong way to build dashboards in tableau.
Relying on grids or overly complex individual charts generally accomplishes two undesirable outcomes in tableau. Firstly, the dashboard doesn’t communicate effectively. Second, is it doesn’t load as quickly as it must.
For example, a sales report displaying 12 months of history for twenty products, 12×20=240 data points, does not help the information consumer see the trends and outliers as easily as a time-series chart of the same information. Also, the quality of the data won’t matter if your dashboard takes five minutes to load. Dashboard viewing is an activity that resembles browsing a website. Web browsing isn’t very useful if you have a slow connection. Viewing a dashboard isn’t efficient, if it takes a long time to load or if the interactivity is slow. The dashboards shown in figure 8.1 displays some common pitfalls –overly dense and complicated charts and inappropriate chart types. Note the pie chart for comparing sales by product sub-category. The stacked bar chart uses a different (conflicting) color legend to display sales by region. The pie chart has too many slices, and performing precise comparisons of each product sub-category is difficult. The cross-tab at the bottom requires that the user scroll to see all the data.
The dashboard fails to convey important information quickly. Presenting the data this way can also lead to performance problems if there are a large number of rows being displayed in the product cross-tab.
Fixing these problems is normally not difficult. The tableau is designed to supply the appropriate graphics by default. Understanding why a dashboard loads slowly and how to ensure good speed requires only a basic understanding of how tableau renders the information. We will dive into those details at the end of this post.
Figure 8.1: A poorly-designed dashboard
|Data Visualization and Dashboarding Fundamentals|