10 guidelines for great Dashboard design
8 minutes read
If you’re into professional cycling, then you might have heard the term ‘Domestique’ before. For those unfamiliar with this strange world of lycra, shaved legs and punishing bike rides, a domestique is a term used to describe a particular role within a cycling team. Like a lot of the terms used in professional cycling ‘domestique’ is a French word and translates as ‘servant’, which pretty much sums up the lot of a domestique rider. Domestiques truly are the unsung heroes of the cycling world. They are the riders who collect the water bottles from the team car and hand them out to the rest of the team. The riders who give their bike or spare wheel to the team leader if they suffer a mechanical problem with the team car nowhere to be seen. The riders who tow a team mate up to the leading bunch and then drop off the pace with the effort once their precious cargo has been safely delivered. Domestiques are the riders who selflessly graft and graft, but never get to win a race, and certainly never get any of the glory and adulation that they deserve.
So what have domestiques got to do with dashboards? Well I think that like domestiques, dashboards are also some of the unsung heroes of the digital world. They too are the grafters. Tirelessly working in the background without so much as a whiff of kudos. Dashboards are becoming increasingly important as the digital age showers us with more and more data. They help to prevent us from drowning in that sea of data, or at least the good ones do. Great dashboards are like windows into a data wonder land. They help users to navigate data, to interpret it, and to unravel the story the data is trying to tell us.
Dashboard design has never been so important, but designing a great dashboard is not as simple as you might think. Here are 10 key guidelines to keep in mind when designing your next digital dashboard.
1. Start with users, not the data
At its essence, what is a dashboard? It’s more than just than a visual representation of data, or just a pretty way to display that data. Stephen Few (author of Information Dashboard Design) defines a dashboard as,
A visual display of the most important information needed to achieve one or more objectives
You see a dashboard, or at least a well-designed dashboard is not about data at all. It’s about information, and about enabling users to use that information to answer questions. Questions like, “How are things doing? What should I be worried about right now? What is the boss going to be asking me about today?”. Without knowing these questions, and without knowing what the important information is, how can you possibly design a great dashboard? You can’t, it’s as simple as that. You can throw data on to a screen in the hope that users will somehow be able to use it to answer their questions, but if you start with the data then it’s highly likely that you’ve already gone off course.
By starting with users you can establish what the most important information is. You can find out what the objectives of the users are and the sorts of questions they’re trying to answer. You can find out the context within which the dashboard is likely to be used, and ultimately you can build a dashboard that is genuinely useful for users.
So what does, starting with users mean in reality? It means speaking to users before you even touch any data. It means asking them what sort of information they need to know; what sort of questions they need to answer and what their ultimate objectives are. It means finding out who the users of a dashboard would be, what makes them tick and what their needs and requirements are. It means finding out where a dashboard would fit in to their wider journey, how they would access a dashboard and possibly even if they’d need a dashboard in the first place. Usually these sorts of questions are best answered via a combination of user interviews and surveys. You should ideally interview a cross-section of users to capture all that lovely qualitative data, and complement it with lashing and lashing of quantitative data via one or more surveys.
2. Don’t flood the user with data
I’ve often thought that a badly designed dashboard is a bit like a ‘Where’s Wally?’ picture (see image below). For those unfamiliar with ‘Where’s Wally?’ (called ‘Where’s Waldo?’ in the US), Wally is a funny looking cartoon character that you have to find in a picture. Ordinarily this would be easy, but these pictures have lots and lots going on. There is a lot of visual noise, making it hard to find little old Wally. You invariably spend 5 minutes trying to find the little fellow and then lose him as soon as you’ve found him. It’s a strangely enjoyable form of mental torture, but not one that you want to apply to the world of dashboard design. Throw too much data at users, too much noise and they won’t be able to find their Wally like nuggets of insight, they’ll just get a headache trying.
When considering the information that you should present to users within a dashboard there is one key questions that you should always ask – Is this information truly useful to users? If it is, then great – stick it in there. If it isn’t don’t show it. You also shouldn’t over-burden the user with unprocessed data. Don’t present the users with all the ingredients that went into making a cake when all they really want to do is eat it.
3. Provide an overview and allow users to drill-down into the data
A great dashboard should be like a quality newspaper. Not in so much as it should be delivered to your door by some pimply teenage paper boy, but that you should be able to pick it up, take a look at the front page and get a pretty good idea of the most important stuff going on. If you want to find out more you can of course dive into the rest of the paper, but just the front page on its own should offer a pretty good overview.
Try to provide users with this front page type overview on the initial dashboard view. Allow users to see the most important information and to drill-down into the data if they want to find out more. Think about how you can boil the data down into its most important parts and how you can use good visual design to present the most important information first.
4. Use visualisations that aid users
Data visualisations are great. If you can use visualisations, such as charts, plots and graphs to help users to interpret and analyse data then you should do. However, don’t go overboard. You don’t want to throw too much at users or data visualise them to death (a particularly grisly way to go).
Think about what is going to be most effective at telling the data’s story. Charts are usually a better idea than pie charts, and certainly avoid over the top visualisations and graphics, such as exploding pie charts and stacked bar charts.
5. Allow users to easily navigate & orientate
It’s likely that a dashboard exists within a wider set of dashboards. Often there will be an overview dashboard, with more detailed views underneath. It’s important that users can not only easily orientate themselves within a dashboard, but also easily navigate between different dashboards. Try to use well know navigation design patterns, such as a left-hand panel, or tabs across the top, and be sure to test these with users to make sure that they are clear. Provide clear labels and signposts so that users know where to go for what, and consider carrying out some card sorting to help test and validate these with users. If you have a hierarchy of dashboards then also consider providing a breadcrumb trail to help users recognise where they currently are.
6. Have a clean, uncluttered dashboard
A good dashboard should be as clean and uncluttered as a Nun’s dormitory room (of course I’m only guessing here, in reality Nuns might be terribly messy lodgers). A good dashboard should not have any unnecessary text, or unnecessary graphics or imagery. You should use good visual design practice, such as white space, alignment and grouping to help visually connect related information and remove any clutter and noise from the dashboard.
7. Use colour to aid data visualisation
Colour is a great way to help users to analyse and interpret data. For example, red, amber and green indicators can be used to communicate a status, and different shades of a colour can be used to indicate higher and lower values. Be sure to be consistent with your use of colour. For example, you don’t want a green indicator meaning one thing in one dashboard, and something completely different in another. Also don’t go overboard with your use of colours – you certainly don’t want a dashboard that looks like it’s been designed by toddler to feature every colour of the rainbow.
8. Provide help and guidance where necessary
Ideally a dashboard should be self-explanatory. You certainly don’t want a dashboard that requires a lot of explanation – that’s always a sure sign of poor design. Often providing some additional help and guidance is important, especially when a dashboard is likely to be rarely used, or frequently used for the first time. Consider including contextual help where it will be useful. For example, provide information to describe a particular data visualisation, or to explain data values shown. To avoid help text cluttering up your dashboard it’s usually best to show it in context upon clicking or hovering over a help icon or link.
9. Use visualisations, language, icons and design patterns that users are familiar with
Try to use visualisations, language, icons and design patterns that users are likely to be familiar with. A novel data visualisation might look great, but is likely to leave users scratching their head. A general rule that I use is that if a well-known standard or pattern exists that users are likely to be familiar with, then you should use it, unless there’s a very good reason not to.
10. Test with users and iterate
This list of 10 guidelines for great dashboard design started with users (quite literally – 1. start with users, not data) and ends with them. This is because without regularly reaching out to users and getting their input and feedback you’re not going to design a great dashboard. When designing a dashboard you should not only start with users, but regularly return to them to explore and validate designs. These could be early designs in the form of wireframes or even sketches, all the way up to full visual mock-ups. You want to answer questions like, do users understand the dashboard? Is there important information missing? Is there information shown that’s not important to users? Most importantly you want to find it if users can use a dashboard to achieve their objectives. This is where usability testing comes in. Ask users to carry out the sorts of tasks they will need to carry out using the dashboard, such as finding out some information, or monitoring the status of something. Capture what they do, what they say, and any usability issues. Test, iterate, test, iterate and repeat – it’s really the only way to build a dashboard that will not only be useful, but usable as well.
- 6 steps to designing better dashboards (InVision blog)
- Designing a Sexy Dashboard (Oleksii Drozdenko)
- Designing a Usable Dashboard (rahulgulati.com)
- 26 Steps of Product & Dashboard Design (Jan Losert)
- 3 Tips for Better Dashboard Design (.prototypr.io)
- Four Cognitive Design Guidelines for Effective Information Dashboards (UX Magazine)
- How to Make an Intuitive Data Display (UX Booth)
GeckoBoard – Marketing dashboard example
Where’s Wally: Fairground by Martin Handford
Dribble analytics dashboard example
Google Analytics dashboard examples