Building apps with Power BI is easier than you think
In this article, I show you how you can take raw data from sources like the Australian Bureau of Statistics (ABS) and turn them into powerful visuals using Power BI. Normally, you might just create an app with a few charts and filters and not think too much about the design. But, if you do this, it could lead to cluttered apps or ones that are not easy to understand/navigate which can lead to users switching off.
Power BI is good at letting you design good backgrounds, creating nice visual elements and making life easier for designers and end-users.
There are plenty of free data sources out there to choose from. In our case, we will take a look at the ABS and they recently revamped their website. In particular, we will look at Labour Force Statistics via an app that I built a few months ago. You can access the latest labour force data here.
In the data download section, we look at tables 1 to 8 for our analysis. Here’s what the data from the ABS typically looks like:
As we can see from the way the data looks, we definitely need to do some clean up before we can do anything with it.
The clean up
There’s an 80/20 rule in many businesses, whether it comes to customers or just something mundane, that is the not so glorious aspect of a job. In the world of data, this is also true and you’ll need to use the Transform Data area in Power BI to do your data cleansing.
Here’s a tutorial from YouTube on how you can navigate the options in this section of Power BI.
Here’s another video on best practices from the well-known YouTube channel Curbal.
For our dataset we needed to do some of this transformation as well to create the tables we want to be able to visualise from:
The data visualisations
For most people, this is the fun part of building things in data analytics tools, but it would be nothing without all the previous work.
When it comes to making these visuals, there are data visualisation concepts to consider and traps to avoid. Here are some things to consider:
- Try to use the right chart – if you’re showing time-series data use a line chart. You can use a bar chart if there are not many points but it can look ridiculous if you’ve got hundreds of bars and no space to show them.
- Keep it simple – too many things going on in a chart can produce clutter and make things confusing – keep it simple with things like colour.
- Make your labels legible – some charts miss y/x axis labels (this may be fine if it is clear what is being shown).
- Keep your colour scheme simple – Power BI defaults to a few basic colour patterns you can change and keep throughout your apps, but if you are highlighting a particular point, it can be best to have that colour stand out and others faded. Also, too many different colours across apps can be confusing, as can having a particular field coloured blue on one page and the same dimension as another colour on a separate page. Consistency helps.
- Consistency – speaking of consistency, this goes with chart design too. Keeping fonts types, sizes, spacing and other factors consistent across chart types, from bar charts to scatter to line charts, is important for readability.
There are good examples on the internet on this topic such as this recent one from the data blog Towards Data Science.
In the article, the author Camila goes through examples of bad looking charts:
And also looks at ways to improve them:
Make sure you consider these things when you build your charts!
Thoughts on UX design and interactivity
When building an app, it’s easy to throw a bunch of charts on a page and menus at the top or sides of your canvas and call it a day. Without a foundation, you might end up with smart apps, but they could lead to nowhere if users find them confusing resulting in them not being used. It is far better to have an app that is simple and functional that allows users to easily navigate than something that has a lot of great insights that are hard to find.
Here is an example of the “throwing a few filters and Power BI charts on a page” approach:
The good thing with Power BI is that it provides you with a pretty cool way to design a background canvas and make navigation menus and buttons that can help guide users through your apps.
To do this we can use PowerPoint (PPT), which is far more powerful than most people think. Firstly, we need to think about a design canvas as you can see below. This is something I created for this app and it is designed by using various block shapes and shadows in PPT.
I create navigations either as buttons that are shaded different colours to indicate the menu option chosen, or plain text that might also be a different colour or made bold to indicate selection. I then create a few of these pages in PPT, and these will serve as backgrounds in Power BI.
The next step is that once you are happy you need to save these down as images. Initially, I’d save them as PNG or JPEGs but the quality was lacking. I eventually discovered that saving them as SVG files resulted in a good quality experience.
The next step is to go into your app for each page that you created and place these SVG images as background images and reduce their transparency, which defaults to 100%.
After this we need to create invisible buttons that you place over the top of the PPT created buttons/menus and then you put page navigation actions on them. This part used to be much more tedious back when the only way to navigate was through bookmarks, so whilst it may be a bit tedious, the world is far better now.
Here is a comparison of moving between tabs on the app versus using the invisible buttons we just created:
Once this is all completed, it’s time to make every page except for the landing page of the app invisible. This way users don’t see menu tabs at the bottom and use the menu you’ve created instead.
Final product and link
If you have a Power BI account this is where you can publish your app to your portal and either share with others in your organisation (if on Power BI PRO licenses or above) or with the wider world (if you’re allowed to share the info).
In our case, I built this using publicly available data so the above app is available here and you can interact with the public version.
Before you go off and do all of this, make sure you read up on and consider the lean startup approach, which is all about building minimum viable product and testing these with users, and then iterating and repeating the process. You do this throughout the cycle of building I mentioned above, because it is a low cost (time and effort and money) way to ensure you are not building something that will not be used.
That said, it’s also important to know, that sometimes you need to show users a better way, an alternative to what they are used to. Yes, the user matters, but if we listened without forward-thinking, we would end up with faster horses, not cars. The same goes for design thinking and technology innovation.
A more fully-fledged example can be seen on this ETF app that I built. It tracks movements in Exchange Traded Funds from the securities exchange ASX and Chi-X.
Our Latest Blog
A MESSAGE FROM OUR CEO | MARCH 2021Right place, right time, right capacity.16 March, 2021 Aiden Heke, Chief Executive Officer, Decision Inc. AustraliaDear Clients, Partners and Staff, Friends ask me ‘how is the new role going?’. Typically, I have three responses.‘I am...
BLOG | DIGITAL TRANSFORMATION In pursuit of progress: defining true transformation 02 March, 2021 Nick Bell, Group Chief Executive Officer “Noun. Movement to an improved or more developed state, or to a forward position.” – Cambridge English Dictionary. Digital...
BLOG | DATA & ANALYTICS 8 Reasons why you need Synapse AnalyticsA single service providing limitless analytics In a recent survey by Harvard Business Review, 80% of respondents said their organisations are struggling to become mature users of data and analytics....