Part 1: Overall Design
Stephen Few describes dashboards as a “visual display of the most important info needed to achieve one or more objectives which fits entirely on a single computer screen and can be monitored at a glance”. Dashboards seem to get more complicated these days and sometimes a single screen won’t cut it. With this in mind, what Few is stressing is the fact that a dashboard should have all the most important information readily available at a glance, as this is really the purpose of a dashboard. If there is a need to display data in more detail, then using drilldown (not available in the current Power BI Release, but for the workaround click here) to bring you to a detail page can be utilized.
Car Dashboard as a Guide
When designing a dashboard, think of a dashboard in your car. Your most important info is available at a glance. Your speed which keeps you from getting a ticket, RPM’s which let you know when to shift gears, your gas level which keeps you from breaking down.
Notice the focus on the most important information to the driver is speed, followed by RPM’s, then gas and oil temp. Everything is readily available at a glance and can quickly be processed by the driver. Indicators such as the gas light contrast to the rest of the dashboard to alert the driver.
All of this is critical information the driver needs at a glance to get to their destination. The drilldown comes in the form of the menu which allows you to click into further detail to display tire inflation and MPG’s, which are nice to know, but not critical to the driver at all times. Keep this in mind in your design. Include only the information that the user needs available at a glance and try to focus on quantitative measures of current data.
Tell a story with your dashboard
The visuals need to inspire action with context, not just show a number. So your sales totals are 520 million this year? Ok, how can we tell if this is good or bad? Is this better than last year, or worse? How does this compare to our forecast? Anyone can look the number up, but it’s the context this number is put into that creates the story your dashboard is trying to tell. If your numbers are down from last year, maybe you changed your sales strategy and it no longer works for you. If your above, you know your new strategy is working.
Effective Graph and Chart Design
Relevancy: Keep the user in mind
When creating a dashboard for a sales department, keep to what they need to know. A salesperson generally will not need a visual representation of inventory levels, but they would need to see measures such as, customer demographics, order amounts, and selling prices etc.
Time frames and Comparative measures
Using time as a measure is very common on dashboards, as it is needed to display trends. To provide context, it is often more effective to use a comparative measure in your chart. If displaying sales this month, you could compare the number to this same month last year. This provides another piece of info that gives the viewer more information to the user on the same visual medium. Instead of having to go and search for those numbers and conduct simple math to find this out, it is available at a single glance.
Precision and Details
Precision is usually a good thing when dealing with data. This is NOT the case with a dashboard. Think of the marks on the speedometer of your car dashboard. The tic marks stand out at every 5 mph interval. The driver needs to know the range. Does it matter if your doing 34 mph vs 33? No. But if you’re over the 35 mph mark in some zones, this can mean a ticket. Same goes with dashboards. If your glancing at sales numbers, would you need to see you are at $4,345,566, or 4.3M. Usually the 4.3M would suffice. If that detail is necessary, utilize a drilldown to give users the option. Furthermore, too much detail clutters the dashboard to make it unreadable. Remember, your dashboard needs to tell a story with the most important info available at a glance.
When showing actuals vs forecasts, keep the forecast as the baseline and the actuals as a percentage of deviation. For example, if you had a planned budget for the year, it would be more effective for the viewers to see the deviation instead of comparing the actual budget number vs. the target number. The baseline would always be zero, and the deviation would be the percentage of deviation the actuals strayed from target.
Readability for Everyone
Assume your viewers may be colorblind. When differentiating objects using color, different shades of the same color usually work the best.
Just Say NO to Pie Charts!
Pie charts can be confusing and display information inaccurately to the viewer. Our goal to provide information accurately at a glance is lost in a pie chart. A slice that represents the highest percentage can look smaller than a slice that represents a smaller percentage and can confuse more than help. A pie chart does not allow for the user to glance and immediately identify comparisons in data, it forces them to try and decipher information that can be better portrayed on a bar graph.
Deciphering the best selling product of 2014 is a bit difficult when glancing at this pie chart. A bar graph would be much easier to read.
Special Effects are Made for Movies, not your Dashboard
Keep the dashboard simple and eliminate distractions. Visual effects such as shading and 3-D are absolutely unnecessary and actually take away from the story your dashboard is telling. They may look cool, but these extras take away from the goal of being able to monitor the dashboard with a glance. The more you add the harder it is to read. Also, stay away from bright colors, unless they are included on key metrics.
One page doesn’t allow for much space to play with, so maximizing it is crucial. Any white space left on a dashboard could have been used to enlarge a chart to make it more readable, or it could be used to add another chart with relevant information.
Keep most important Data on the Top Left Corner of the Screen
This ones pretty self explanatory, but still overlooked. A users eyes tend to wander to the top left, as that’s where we are trained to start reading. Utilize this human habit in your dashboard and always put the key metrics in the top left corner.
Minimize Text Within Visualizations
A dashboard uses visualizations for the purpose of giving immediate insight into the story it’s trying to tell. The more text, the longer a user must look at the dashboard to decipher information. Too much detail takes away the effectiveness of a dashboard. This is not to mean the text isn’t to be used, in fact, text is sometimes the best way to communicate data.
If you need to show only YTD Sales with nothing to compare, just leave it at that:
Chartjunk belongs in the trash
Edward Tuft coined this phrase to describe unnecessary pictures and objects that only serve the purpose of decoration. Pictures and logos take up valuable real estate and are useless to the user.