Power BI Best Practices: 10 Guidelines to Always Follow

This one is so simple yet always overlooked.

When designing a dashboard or a report, always consider your audience and how they would view the canvas.

In general, people read left to right, top to bottom.

Expectedly, you should only pin visuals that contain important data.

However, some of them may be more important than others.

Putting them on the top left corner is a great way to be seen first.

Here are ideas to accentuate your most important information:

• Increase the text size
• Make the visual’s color stand out
• Use the “Card” visual

For example, on this dashboard, it’s pretty easy to spot the data about the total number of stores and new stores:

Whatever method you use, the main point is to make the most important data stand-out to your users.

In general, visuals look pretty and stunning in Power BI.

However, looking good isn’t the same as effective.

Don’t just use a visual just to have different types of visuals. Make sure your visuals tell a story and are easy to read and understand.

Here are some guidelines and ideas in using the right visuals:

• Be wary of circular chart types. If you need to use one, make sure you have 8 or fewer categories.
• 3-D charts look good but they are harder to read than flat visuals.
• Be consistent with your designs — chart scales, dimensions, time frames, measurement scale, and even colors.
• Use only 3-4 numeric values when showing numbers — 1.34 million is better than 1,340,000.
• Use only data labels when necessary.

Remember these guidelines even when using Power BI’s best visuals.

One potential problem with using Power BI is being overwhelmed with the visuals you can use.

Creating a Power BI report with too many visuals could slow its performance. And that’s inconvenient for both you and your users.

Like this:

Follow the 8-10 rule:

• When creating reports, use a maximum of 8 visuals per report page.
• For dashboards, use a maximum of 10 tiles.

In simple terms, RLS restricts data row access of a user based on his role.

One of the benefits of RLS is that it makes the design and coding of your application’s security simpler and easier.

When RLS is enabled, Power BI would only take and import the data the user is authorized to view.

When it comes to RLS, it’s recommended that you:

• Create a separate schema for the objects, predicate functions, and security policy
• Avoid type conversions
• Avoid recursion in predicate functions when possible
• Avoid using excessive table joins in predicate functions

To do this, all you need to do is define the roles and rules of your users.

Filters and slicers are similar but not exactly the same.

Let’s start with Power BI filters.

Filters help you show selected data to users on different levels — visual, page, or report.

It’s a great feature. But here’s the problem:

The larger the data the visual needs to display, the slow it will take to load it. The more memory Power BI uses, the more the performance declines.

To compress some of this data, filter the visuals shown when possible.

On the other hand, slicers are different.

Though Power BI slicers are visual filters, each use by the user generates 2 queries:

• Get the data
• Get the selection details

Too many slicers could affect performance in a negative way.

Power BI has lots of visuals — but not everything.

An example of this is a histogram. Power BI doesn’t have a built-in histogram chart.

The good news:

There’s a marketplace that has custom visuals made by 3rd-party individuals and companies.

The bad news:

Uncertified custom visuals may contain risks (data leakage, malicious data injection, etc.) and may not perform that well.

All the visuals in Power BI are interactive.

If you try to click on an element on a certain chart, the other charts on the report page would respond and highlight data related to that element you just clicked.

However, these interactions have a negative effect on performance.

To optimize your reports’ performance, control and disable unwanted interactions between the visuals.

If you’re not sure how to do it, the simplest way is in Power BI Desktop:

Click the visual, then ‘Format’ on the tab list, and select ‘Edit interactions’.

This shouldn’t come as a surprise.

When importing files, especially workbooks, only import fields and tables you will be using.

For example, in Spreadsheeto’s Power BI histogram tutorial, we have prepared a sample Excel workbook to import.

In there, our instruction was to only import the table which contains the data.

The bottom line is, as much as possible, keep your data model as clean, narrow, and lean as you can.

After importing only the important fields and tables, it’s time to optimize your data model.

Here are some of the things you can do:

• When dealing with numeric data, use integers instead of strings.
• Check your DAX functions. Some of them could increase memory requirements and run-time which in turn, slows your report.
• Be specific as possible in your columns and avoid fields with unnecessary precision and high cardinality.
• Index columns that are usually filtered or sliced when connecting with other data sources using DirectQuery. This improves report responsiveness.

If you’re interested to learn more about the last item above, Microsoft’s Analysis Services Team prepared a nice whitepaper about optimizing data sources for DirectQuery.

Though Power BI helps find insights about data, a lot of users count on Power BI’s ability to tell a story about the data to a non-technical audience.