Power BI allows you to start producing reports and dashboards immediately, thanks to its intuitive and simple operation. Mastering it, however, requires more in-depth knowledge and time investment. Nonetheless, for those starting out, there are some Power BI Best Practices that can help you take the first steps while avoiding the most common pitfalls.
- Use reliable and accurate data sources. To get insights you can trust, make sure that the sources you use contain reliable and accurate data that match what you need to show. Otherwise, it’s like starting the process with an asset that will contaminate the end result.
- Consider who your audience is. Map your audience before choosing sources and indicators. It is not the same to build an internal report as doing it for a partner or a customer. Each will be looking for a different side of the information.
- Choose your KPIs carefully. Think carefully about the Key Performance Indicators you need to work on to deliver the big picture to your audience. What will they be interested in? What level of detail is appropriate? Normally, you will tap onto enormous, huge sets of data. But remember, you do not need it all: just a few handpicked fruits.
- Do not display private or confidential data. Be careful with the data you include and make sure everybody in your audience has been authorized to see it. For example, confidential information on profit margins may be only for people within the organizations, and not for external partners. Double-check you are not displaying sensitive info to an audience not cleared to read it.
- Choose appropriate visuals. Bar graphs are specifically created to display multiple rather than single instances of a measure. Another good choice for categorical data is the stacked chart. For time series, you can use lines or charts like Waterfall and Combo. When you are working to show relationships between different measures, the scatterplot chart is what you should be looking for.
- Be functional. Decoration is fine, but never at the expense of functionality. A data visualization report should only be beautiful when its beauty can boost understanding in some way without undermining it.
- Be familiar. It would be best if you found a correct balance between originality and familiarity. Designing an original report is OK, but you also need to work on familiarity, using recurrent patterns to make the user experience a more comfortable one, an original layout but with recognizable patterns across the pages.
- Keep yourself updated. Power BI is constantly evolving, both desktop and online versions (especially the latter.) Every now and then, new functionalities are added, or existing ones are improved towards obtaining quality insights that can be put into practice. Keep up the pace!
- Get professional training. An interesting option to walk through the basics is to join a Dashboard in a Day: workshops organized by Microsoft and its partners to teach the basic handling and possibilities of Power BI in a single day. Additionally, many companies offer courses with the assistance of experienced mentors.
- Learn from the community. It is easy to find templates with predesigned dashboards and reports, webinars, and other learning resources. Many are available on Microsoft’s website. Also, the Power BI community is incredibly active, and every day shares best practices or challenges well-treaded paths in search of new solutions. Look for interesting Power BI blogs and check them frequently, it can work wonders!
- Power Query + DAX + Visualization. Success depends on your ability in the three pillars that support Power BI. If your PowerQuery is poor, your data will be dirty or not optimized for your target. If your model does not respect modelization best practices, star schema and each DAX formula will become an Everest-like climb. If your visualizations are bad or too complicated to understand, users will abandon them. Work all skills in parallel.
- Choose the appropriate type of data visualization. First, we should determine what the user intends to do with our data, to choose which type fits better. A dashboard is a one-shot visualization. The idea behind this is that you instantly apprehend all you need to understand. Reports, instead, are interactive tools designed to drill data at multiple levels. When projecting a report, you must prepare the user to travel through the data. Finally, presentations are where data storytelling skills can shine and show their full capabilities. You can use more artifacts and tricks than in the reports.
- Pie charts – beware! You will almost certainly include a pie chart somewhere. Most people do. However, pie charts can be confusing at first glance when there are more than three categories, due to the number of colors, labels to include, etc. They are more appropriate when a part needs to be highlighted against the whole.
- Use your colors wisely. Colors should be used consistently to provide a clear interpretation. Be sure to use the same color for the same category in all the visual objects you use in a report, throughout. For example, if you use blue to indicate a positive magnitude and black for a negative, do not reverse the meaning in the following graphic. Check with the marketing department what color palette your organization uses and align your work with it.
- Tables and matrices. In 95% of the cases, adding tables and matrices will cause your report to become confusing and not achieve its goal of communicating clearly. Instead of cluttering a report with tables, let your users explore the data in the charts or the model itself, through Excel. And, if you cannot avoid using a table, try not to have too many rows or columns, so that the most relevant information can be quickly apprehended.
- Custom visuals. In the Power BI Marketplace, you have at your fingertips more than a hundred custom visuals. Keep this in mind: what features do they offer you that the existing visuals will not provide? Some customs objects do not support all Power BI features, such as data exploration or labels. Also, be careful because updates may take longer to arrive…
- Eliminate the superfluous. Try to include only the most relevant information and do not be afraid to discard that which does not offer maximum value. Simple and clear, that should be your ideal. In a way, it’s like a joke: if you have to explain it, it’s not funny.
- Experiment. Don’t settle for the first solution that comes into your head, try different alternatives, look for optimization, that offers you more with less. You will probably have to play a bit with the piece, but you will arrive at a better result and gain skill through practicing!