Dashboard Design - Standards, Principles, & Best Practices
Nick Hilton | 18 August 22 | 8 min read
Welcome to this edition of Infused with Data. As someone with a focus on modern design principles and visualisations this article on dashboard design has probably been my favourite to write so far. As well as being fun to write it has been helpful for my learning and has things I'll take away from it to take my reports to the next level, I even snuck a Simpsons reference in there! I hope you find it helpful too. Let's get started...
Whether you’ve just started out making reports in Power BI or have been designing BI solutions since before the term ‘business intelligence’ even existed you must have a point in every new report where you look at the design process and how you can most effectively achieve the objectives of the reporting.
If you’re new to this or are a report designer and don’t engage in design planning, then I highly recommend investing some time in these fundamentals. That’s what these are, fundamentals, and if you don’t make them a part of your toolset sooner or later, you’re going to find the report you just developed isn’t received anywhere near as well as you’d hope. Unfortunately, this is one of those things in life where if you do it well it probably won’t be noticed or commented on at all but that’s often a part in producing a good piece of work.
In this post I’m going to go over a few tips and concepts which I hope will help you out in a big way next time you’re designing a report or reporting solution
This will be split into five easy sections with each tip being to the point and digestible:
-
Your First Steps
-
Accessibility
-
Narrative Efficiency & Effective Data Storytelling
-
Fixed v Dynamic Design
-
Helpful Resources
Your First Steps (Yes, every time!)
Understand the business, understand the data, understand the requirements - The first step on any data journey is to invest some time understand the business and its (relevant) processes. By focusing on processes rather than departments or functions consistent information is delivered throughout the organisation. This is the real world and the real processes which create and dictate the flow of organisational data. The next steps are to develop the scope of the requirements and to really review the data (make some notes!). At this point you should be able to get a good grasp on the entirety of what you’re working with.
Keep it simple! - Charts should be easy to glance at and understand. No squinting. You want to provide data that’s easily digestible.
Get familiar with and stay up to date with design practices and trends - I’m sorry to say that UX and software design changes almost as quickly as the weather does so your years of experience or background in report creation may be easily surpassed by someone eager to stay on top of the latest trends. I advise that anyone who frequently develops reporting solutions should be reading UX design reports at least once every 12 months. There are all sorts of bubbled edges and overblown shadow effects which looked great 5 years ago but now look a bit rubbish. This is nothing to do with talent or your eye for design it’s about changing externalities.
Accessibility
Specific end user requirements - Once upon a time I started a business to create resources for GCSE pupils which utilised modern technology rather than the same old textbooks. In one eye opening discussion I learned that at least 10% of the population suffer from conditions which make traditional textbooks and reports difficult to make use of, conditions such as dyslexia, dyspraxia, or colour blindness. Look, these are private details so be mindful you may not want to ask about this outright, but a few carefully composed questions could be the difference between your report being illegible or being highly effective. Think how easy it could be to fill your report with gorgeous conditional formatting only to find out the CEO reading your report can’t even see those colours. Below is a tool built into Adobe Color which provides accessible colour mixes at the click of a button.
Choose a colour palette - Most of the time we’re fortunate to be able to start with at least one colour requirement. Something like a primary brand colour. Chances are that you’re building full reports and visualisations so you’re going to need to know what complements and works well that colour. Amazingly, there are tools from Canva and Adobe (and probably many more) that are totally free to use, will take any image, and provide a pallet instantly. Here’s Adobe’s version
Check for brand guidelines - During the design stage investigate any branding requirements, often companies will have documents with colour samples and imagery you can use.
Be mindful of page space - We often try to fit too much onto one page, after all there may be 10 different Sales metrics and one perfect tab for all those metrics but there’s a very fine line between not enough information and too much. That sweet spot is exactly what to aim for. You want the page to be full but not cluttered and for there to be reasonable amounts of space between each element. Effective use of that canvas is a skill but one that’s easy to develop through practice. At the bottom of this page I’ve left some links to examples of great dashboards after that there is a world of information on how to do this well. (Power BI users specifically I recommend looking into bookmarks as a way to layer your reports, it looks cool and means you can fit a lot more in).
Narrative Efficiency & Effective Data Storytelling
What is data storytelling?
“Data storytelling is the ability to effectively communicate insights from a dataset using narratives and visualizations. It can be used to put data insights into context for and inspire action from your audience.” (C.Cote, Harvard Business School, 2021)
I can’t put it better than that! There are two aspects of this which you will develop a) The skill of building reports which effectively communicate insights, and b) The skill of being able to literally communicate your insights, in a meeting for example.
Don’t forget the business objectives - All the design in the world won’t save your report if you forget to align these principles with the business requirements and processes. Ask yourself throughout how well the report gets users to where they need to be, can they now take action? If this is a report meant for executive meetings consider whether your report will generate discussion amongst a group rather than encourage deeper and deeper investigation by analysts.
Make sure you are able to choose the best visualisation - Spend the wildest evening of your life going through the most commonly deployed data visualisations, if you’re using Power BI then go through each of those. There’s a reason each one exists, because at some point they are the best way to effectively communicate the insights. This site is a brilliant resource.
Avoid reports that drag on - If possible, try not to have reports that rival Steven King novels in number of pages. Often less defined reporting requirements (which is natural) can lead to sprawling reports which seemingly never end and at some point either the report or most of it’s pages will become redundant. Occasionally reports will end up this way and have requirements for that but avoiding unnecessary bloat in any case is important.
Be consistent - Make sure your design, terminology, formatting, and everything else is as consistent as it possibly can be. Reporting should flow and not feel disjointed. The key here is to do this at the level of the report users as you’ll often find many organisations’ own branding or internal terminologies may not make much sense. For example, one term may mean two totally different things in two contexts one in Sales and one Finance. You need to get your head round these things of course but what matters is that it’s designed with the end users in mind.
Fixed v Dynamic Design
Build your dashboard using a modular design - Modular design is dynamic design. Creating modules you can add, reconfigure, edit, and subtract over time means that your dashboard stays relevant to your users’ adapting needs.
Build a data model that can scale - How much impact you have on this depends on your involvement in the ETL at the database layers or the modelling in Power BI. Whichever the case if you’re making reports which may grow to serve more users then make sure you are aware of ways to build data models in a way which can scale.
Use a golden data set - This means to keep your data model separate from the actual report. So in Power BI you build your data model and have a file for everything like measures, you publish that and then create a new PBIX where you choose the data source from the Power BI service. This way you can minimise file sizes, swap models (dev and live) more easily, and potentially most important nowadays you can securely work on and distribute the report files without risking all the data sitting in the model to the same degree.
Keep your data model tidy - Sometimes reports are going to end up with hundreds of measures or columns, you can’t always avoid that. What you can control is whether all of those things are neatly sorted into folders like ‘Finance measures’ or whether ‘ColUMNs aRe LaBELleD LiKE This’. By sorting this out you’re making life 100x easier for yourself and then of course making sure you’re delivering a piece of work that your client or users could work with themselves if needed to.
I really hope that this list of tips comes in useful, don’t worry about learning them all just slowly integrate those that stand out into your work and you’ll naturally grow and push your standards as a report designer. And of course if you need any further advice or guidance please don't hesitate to get in touch with us.
Helpful Resources
Here are some helpful resources, tools, and articles I’ve found along the way:
Guy in a Cube - The #1 Power BI YouTube channel
Curbal - Another top Power BI YouTube channel
Dave McCandless - You might have seen his books; this guy presents data in the most amazing ways.
The Data Warehouse Toolkit by Ralph Kimball - The #1 book if you’re looking at the ETL side of things