In the last section of this chapter, we want to discuss the challenges that you face when creating high-density reports. When you include many visualizations in a single report, you need to carefully balance the amount of information provided in each of those visualizations, removing all the unnecessary elements that would reduce the attention of the user. You want your user to focus only on the data.
145 CHAPTER 7 | Improving Power BI reports
As you will see, having an idea of the overall structure and tuning properties of each visualization is much more important than using particular custom visualizations, which are usually the icing on the cake.
Figure 7-32 depicts a first version of a report showing website data from Google Analytics for the DAX Formatter website, which is available in the companion content, in the file Sample-DAXFormatter-Analytics.pbix file. The report contains 28 visualizations with data, plus one slicer and eight components without data that are there only for aesthetic reasons (title, logo, pictures, and separators). The 28 visualizations only use 7 different visualization types, and some of them are simple variations of the same concept, such as stacked/clustered bar/column charts. You can obtain a complete and complex report using only a few component types.
Figure 7-32: A high-density report based on Google Analytics data.
The entire report is organized in three zones: left, center, and right. The left zone contains metrics regarding the number of users, the center zone shows data about the sessions, and the right zone includes technical information about average page load time, device type, operating system, browser, and resolution used by website visitors.
If you were to try to create a similar report starting from scratch, you would need to apply the following guidelines:
Reduce text Include only the minimum necessary of textual elements, avoiding repetitive or verbose descriptions.
Remove legends Whenever possible, avoid including a legend, especially whenever there is only one measure displayed in the chart.
146 CHAPTER 7 | Improving Power BI reports
Remove axes In a compact visualization for which you already included the data labels (setting the Data Labels property to On), you can remove corresponding axes. All the clustered bar charts in the report are formatted in this way.
Use images to explain concepts Use an icon or a meaningful image related to the data you show. Remember, a picture is worth a thousand words. In the example in Figure 7-32, at the top of the report is one different image for each of the three zones (Users, Sessions, and Average Page Load Time).
In the report, we used a donut chart. Previously in this chapter, we mentioned that using a pie chart or a donut chart is not a good idea, so you should avoid doing that. The exception we wanted to include in this report is when you compare only two values. In this example, we used a donut chart for the search engines distribution, showing the percentage of sessions coming from Bing searches versus Google searches. It is clear that Google has a clear leadership for directing visitors to this website, with Bing producing only a marginal contribution. For this difference, looking at the exact number or percentage is not relevant.
There are three visualizations that we wanted to improve, starting from this initial example. The sparklines used at the top of the report are simple line charts without any axes, legend, data labels, or border. Figure 7-33 illustrates that we set to Off most of the format properties of those line charts.
Figure 7-33: The Properties pane of a line chart used as a sparkline.
However, you cannot change the line width of the line chart. When you use the line chart in a small area, this produces a result that is difficult to read. You can replace the line chart with the Sparkline custom visualization that you can download from the Power BI visualization gallery. Figure 7-34 shows a side-by-side comparison of the same chart displayed by using a line chart (on the left) and a
sparkline custom visualization (on the right). The custom visualization draws a line with a smaller width, generating a final result that is more readable.
147 CHAPTER 7 | Improving Power BI reports
Figure 7-34: A couple of examples of the same chart, using a standard line chart and a sparkline custom visualization.
Another visualization to improve is the countries/regions penetration displayed in the lower-left corner of the report. Instead of using a standard map, which shows a pie for each country/region with a size depending on the population density, you can use the Synoptic Designs panel, loading the world map from the gallery shown back in Figure 7-24. The result of displaying the measure Users Per Million in a Synoptic Designs pane is visible in Figure 7-35.
Figure 7-35: Synoptic Designs panel with a world map that displays the ratio between users and population.
The last visualization to improve is the one in the lower-right corner, showing the number of sessions by browser resolution. In the original data, we have a very fragmented number of different
resolutions, so the bar chart only displays the first values, but the number of sessions of the most common resolution is just five percent of the total number of sessions, so there are a wide number of different resolutions considered, most of them not visible in the report. We solved this problem in two steps. First, we created a column containing the width category, which classifies the width extracted from the resolution string, as you have seen in the previous section about DAX. Then, we changed the clustered bar chart to a waterfall chart, using the Sessions % measure instead of the Sessions measure, as shown in Figure 7-36.
148 CHAPTER 7 | Improving Power BI reports
Figure 7-36: A waterfall chart with the distribution of sessions by browser resolution.
We do not use any decreasing step in the waterfall chart, but the final result clearly shows the distribution of the resolution in a meaningful way. The Sessions % measure is a DAX expression created just for this report, using the following definition:
Sessions % = DIVIDE (
SUM ( Website[Sessions] ), CALCULATE (
SUM ( Website[Sessions] ), ALL ( Website[Width Category] ) )
)
Figure 7-37 presents the final result, after these improvements have been incorporated. As you can see, the overall difference is an incremental improvement and not a substantial change from the result we got initially by using standard visualization components.
149 CHAPTER 7 | Improving Power BI reports Figure 7-37: A high-density report based on Google Analytics data.
Especially in a high-density report, you should focus on the overall quality and readability, reducing the number of distractions for the reader. There is already an overwhelming amount of information, so the focus of the reader should be entirely on data, not decorations or visualizations that are too complex and do not provide any added value.
Conclusions
In this chapter, you have seen a number of techniques to improve Power BI reports by choosing the best built-in visualizations and adding custom visualizations. Here are the main steps in this process:
Choose the right visualization type You can choose between many visualization types, but usually you do not need too many of them in the same report. Do not be afraid of using the same visualization type many times if it is the one that best displays your data.
Customize visualization properties You can customize every visualization by using format properties. Using a consistent color scheme is one of the most important aspects of a good report.
Consider custom visualizations when necessary The Power BI custom visualizations gallery provides you with many visualizations that extend the set of standard ones available in Power BI.
You should consider using them when there is a concrete advantage over the standard visualizations.
150 CHAPTER 7 | Improving Power BI reports
Use DAX to create measures and calculated columns You also can use DAX expressions to achieve the desired visualization. Even a simple transformation, such as renaming a measure in a legend, is not always possible within the properties of a visualization, but you can overcome this by creating new measures and calculated columns.
Remove unnecessary elements in high density reports In a high-density report, you need to remove any graphical element that is not necessary to communicate information to the users; you do not want to distract them by including details that do not provide any useful information.
These guidelines are just a starting point in your journey to create clear and useful reports. Your experience, the feedback from users of your reports, and the analysis of reports created by other people are the other important steps along this road.
151 CHAPTER 8 | Using Microsoft Power BI in your company
C H A P T E R