It is best practice to narrow the focus of your dashboard to a central premise. Dashboards that attempt to address a wide variety of stories or premises often become overly complex. In general, if you end up needing to create help or user guides for your dashboard it may lack the focus required to be optimally effective.
|Interested in learning Advanced Tableau? Click here to start learning more in this "Tableau Training and Certification Course" at Mindmajix.|
There are a number of ways Tableau facilitates telling your story, and there are a number of themes you can take advantage of when using Tableau. It is common to have several different methodologies presented in a single workbook, but it’s more intuitive for end-users when a single dashboard supports the story with a single methodology. Some of the most frequent methodologies are:
1. Geography-centred dashboards: One or two choropleth maps (i.e., filled maps) with ancillary supporting elements to refine your story as it relates to geography.
2. Time-centred dashboards: One or two key charts with a primary dimension being time. Some time-centric dashboards may also include a time-based paging element, frequently including historical values, for animating the dashboard over time.
3. Guided analytics: Sequence-specific interactive elements that build/compound on each other. Be careful when using this approach as end-users can get lost if your elements result in circular filtering (i.e., element A filters element B, which filters element C, which filters element A).
4. Ad-hoc investigation: The central element(s) vary quite a lot, but most ad-hoc dashboards will provide a variety of parameters and quick filters to allow the end-user to “tweak” or refine their path of investigation. Ad-hoc dashboards generally, cannot predict every possible investigative path end-users might take, but a number of key paths should be supported with a variety of possible add-on filters for each.
After you have analyzed some data and determined what information you need to share, adhering to these principles will help you create better dashboard designs:
These principles come from personal lessons learned during building dashboards in a wide variety of use cases. They work well for 90 percent of the use cases across the industry, government, and education.
You may find specific use cases for which violating one or more of these best practices perform well and communicates the information effectively. By all means, then, do what works best for your specific case.
Dashboard building would be easier if everyone had the best computer with high-resolution graphics. Unfortunately, this normally isn’t the case. So, you must design your dashboard to fit comfortably in the available space by determining the pixel height and width of the worst-case dashboard consumption environment.
Tableau provides defaults for the typical sizes you will need or allows you to define a custom size. Doing a lot of design work without knowing the consumption environment is a recipe that results in unhappy information consumers and extra work for the designer.
|Checkout Tableau Interview Questions|
Will the dashboard be consumed on laptops via tableau reader? If yes, do you know the range of screen resolutions that are being used? Are tablet computers used? Is the dashboard going to be consumed via the tableau server or will you have to embed the dashboard in a website? You need to understand the specific height and width of your dashboard space.
For laptop consumption, this can be as little as (800×600) pixels. For desktop computers or better resolution laptop monitors, (1000×800) pixels normally work well. Web embedded dashboards can be smaller but a typical worst-case minimum size might be as little as (420×420) pixels. Tableau has predefined sizes to help you layout the dimensions of your dashboard. Tableau also makes it easy to define custom size ranges, if the default values don’t meet your needs.
Four individual visualizations will fit well on most laptop and desktop computer screens, as shown in figure 8.3. This style of presentation naturally highlights the upper-left pane because people of western societies have been taught to read from the upper left to the lower right of a page. Figure 8.3 shows a 4-panel design intended for laptop or desktop consumption.
A four visualization design style will generally be read from the upper-left to the lower-right in a Z pattern unless you do something to grab attention elsewhere. Note that design actually includes five panes-but the fifth pane, (the small select year cross-tab) acts as a filter for the rest of the dashboard. Ordinarily, a quick filter would be used to permit the audience to select the year in view.
Instead, the example in figure 8.3 uses a small crosstab to trigger a filter action, the advantage of using a crosstab instead of a quick filter is that additional information is provided (total sales for each year) in the same amount of space a multi-select quick filter would have required. The design employs a fifth data pane (an apparent contradiction to the best practice) but in a way that is consistent with the recommendation. Another reason to use a crosstab for this purpose leads to the next best practice recommendation.
|Related Page - How to build your first advanced dashboard in tableau|
Different rules apply when designing dashboards for tablet computers. Designing for tablet computers will be covered in detail at the end of this post.
Figure 8.3: A 4-panel design
Using actions in place of quick filters provides a number of benefits. First, the dashboard will load more quickly. In order to visualize quick filters, tableau must scan the source table from your database. If the table you are scanning is large, it can take some time for tableau to render the filter. Tableau has improved quick filter load performance over the last several releases, but you may opt to use filter actions for another reason-aesthetics.
In the same space that is required to display a multi-select filter, you can provide a small visualization with a filter action that enables filtering, but in a way that also enhances the content included in the dashboard.
Employing multiple quick filters in a dashboard is also potentially confusing to the audience. My personal worst-case scenario involved a client dashboard with two data panes and thirteen quick filters. The source database was very large (billions of records). It required six minutes and thirty seconds to load-all but eight seconds of that time were required to visualize the quick filters. Not only was it difficult to find the right filters, but it was also slow loading.
By altering the design to a series of 4-panel dashboards and replacing the quick filters with filter actions, the load time for each dashboard was reduced to less than eight seconds. This leads to the next best practice recommendation.
Achieving fast load times can be challenging if the source data is very large. In the case mentioned in the preceding section, the load speed of the dashboard was terrible because many of the thirteen quick filters were scanning massive tables.
The executive that requested the dashboard needed to be able to have data summarized globally, but also wanted to be able to drill into much more detailed subsets of the data. Unfortunately, the initial design was slow-loading and didn’t provide much insight. By creating a series of four-pane dashboards the load speed was improved dramatically and the understandability of the information presented was greatly enhanced.
|Related Page - What is the right way to build a dashboard in tableau|
The redesigned primary dashboard provided a good overview of operations by showing a bar chart (comparing different products), and a map (to show data geographically), a scatter plot (to provide outlier analysis), and a small crosstab with very high-level metrics. Filter actions were added to these visualizations that allowed the executive to see more detailed information in other dashboards that were pre-filtered by the selections made on the main dashboard. This cascading dashboard-style provided all the information requested, but in a way that improved load speed and understandability.
The final design replaced the original dashboard (containing two data-objects and thirteen quick filters) with four cascading, four-panel dashboards. The top-level dashboard provided a summary view but included filter actions in each of the visualizations that allowed the executive to see data from different regions, products and sales teams. None of the new dashboards required more than eight seconds to load.
If you employ this recommendation and you are experiencing slow performance, tableau’s performance recorder provides visibility of the technical details you will need to troubleshoot the issues that may be degrading the performance. The desktop performance recorder is covered at the end of “bringing it all together with dashboards” post.
Too much color on a dashboard is confusing. Try to limit the use of the color to expressing one dimension or one measure. You can effectively add a secondary use of color in the same dashboard if that secondary use of color employs a more muted color scheme.
The dashboard in figure 8.2 used two colors more effectively than the dashboard in figure 8.1 because the secondary use of color expressed a limited set of values (true/false) and the color was expressed using a muted shade of gray.
According to data visualization expert and author Stephen few, up to ten per cent of males and one per cent of females have some form of colour blindness. The most prevalent form of color blindness limits the ability to distinguish between red and green. Take this into consideration if your dashboard will be consumed by a large population.
To avoid potential problems apply grayscale or blue-orange color pallets. These are visible to most color-blind people. Tableau also provides a color-blind pallet with ten colors. You may also consider building color-blind specific dashboards if you have a very large population of information consumers.
Quick filters are obvious. Actions are not. Since actions are triggered by selecting elements of your visualizations they will not be obvious to your audience unless you provide instructions within the dashboard. Placing instructions in the title bar of the worksheet that triggers the action is a good way to remind people about the availability of the action.
Use a consistent font style and color for these instructions in your dashboards so that your audience learns that style denotes an instruction. The instructions used in the figure 8.2 dashboard are highlighted through the use of a brown italic font.
|Learn Tableau Tutorial|
Another alternative is to place the instructions in tooltips that appear when hovering over marks, as shown in figure 8.4. This method offers the advantage of having the instructions appear in more complete text without crowding the dashboard space.
Note that the format of the instruction matches the color, font, and style of the instruction in the headers of the dashboard.
Give your audience even more explanatory information by adding a separate read-me dashboard that includes additional details regarding the data sources used, formulas used, and navigation tips. You can include links to websites that provide even more information, as shown in figure 8.5
Figure 8.4: Instruction in a tooltip
Figure 8.5: A read me dashboard
Finally, provide your contact information so that people can easily ask any other questions that may not be anticipated in your design.
Crosstabs are useful visualizations for looking up specific values when you know exactly what you’re looking for. Crosstabs are not the best visual style for quickly discovering trends and outliers. Figure 8.6 shows the poor use of a crosstab view. Even though the crosstab in view has been filtered for a specific dimension, vertical scrolling is still necessary in order to see all of the state values.
There is also a lot of white space generated by the column headers for market and state in figure 8.6, this dashboard could be improved by creating a filter action from the bar chart that could restrict the market displayed in the bar chart to a single market, but even with a filter action for the market, the crosstab would still require scrolling to see all the values.
Figure 8.6: A poor use of a cross tab
In figure 8.7, the crosstab is much more compact. The market and state dimensions are being displayed in the title dynamically and the orientation of the crosstab has been changed to place the measures (11 fields) in rows and the product type (4 fields) in columns. This reduces white space and eliminates the need for scrolling.
Figure 8.7: A better use of a crosstab
The unfiltered version of the dashboard on the left of figure 8.7 clearly shows all the information without any scrolling. The filtered version on the right of figure 8.7 shows more granular data in both the time series and the crosstab. This is accomplished using filter actions triggered by the bar chart and map-providing details on the demand for the markets and states of interest. The use of dynamic titles in the time series and crosstab visualization (highlighted in figure 8.7) communicates the information more effectively in less space.
|Related Article - Building Interactive Tableau Dashboards|
This best practice rule was inspired by Edward Tufte, author of the visual display of quantitative information. Remove any text, lines, or shading that doesn’t provide actionable information. Remove redundant facts. If a company logo isn’t required for promotion purposes, then remove it. Ruthlessly eliminate anything that doesn’t help your audience understand the story contained in the data.
Trying to save time by making one dashboard serve many purposes will not result in the best-performing dashboard or save design time. It is so easy to build dashboards and apply data restrictions within data extracts that I recommend making your dashboards fit the particular purpose of each audience.
Generally, executives need to see high-level data across multiple geographies, product lines, and markets. Regional staff needs more granular data, but for restricted geographies, products, and consumers.
While it is possible to make one dashboard that works for both groups, it normally doesn’t produce the best possible format or the best performing dashboard for either. Strive to provide the best possible experience for each audience even if that requires a little extra effort.
Achieving fast load times are dependent on the size and complexity of your data as well as the type of data source you are using. Slow loading dashboards can also be caused by poor dashboard design. There are several ways that the dashboard design itself can contribute to slow load speeds.
Including high granular visualizations (that plot a large number of marks) can consume resources and cause slow load times. Using too many quick filters or trying to filter a very large dimension set can slow the load time because tableau must scan the data to build the filters.
Tableau includes built-in tools for both tableau desktop and tableau servers that help you identify performance issues. At the end of this post, you’ll learn about the desktop version of tableau’s performance recorder. The server version will be covered in the “installing tableau server” post.
|Related Page - Learn What is the wrong way to build a dashboard in tableau|
List Of Tableau Courses:
|Data Visualization and Dashboarding Fundamentals|
|Tableau Server Tutorial|
Stay updated with our newsletter, packed with Tutorials, Interview Questions, How-to's, Tips & Tricks, Latest Trends & Updates, and more ➤ Straight to your inbox!
|Tableau Training||Feb 07 to Feb 22|
|Tableau Training||Feb 11 to Feb 26|
|Tableau Training||Feb 14 to Mar 01|
|Tableau Training||Feb 18 to Mar 05|
Viswanath is a passionate content writer of Mindmajix. He has expertise in Trending Domains like Data Science, Artificial Intelligence, Machine Learning, Blockchain, etc. His articles help the learners to get insights about the Domain. You can reach him on Linkedin
Copyright © 2013 - 2023 MindMajix Technologies