Dashboards That Make You Feel Warm & Fuzzy

P Van
Adventures in Data, Design & Development
6 min readSep 26, 2014

--

“Dashboard”

What do you think of when you hear that word? I think of horrible, awful 3D pie charts. I think of light blue area charts, a la Google Analytics.

I see jagged, unfamiliar landscapes. I see a few big numbers. The cool, calm blue hues make me feel relaxed. But would I still feel so chill if I understood what was going on in the data?

This dashboard is not asking me to “do this!” or “maybe do that!” This dashboard is not telling me “Things are chill in this area man” or “Oh crap you need to manage this!” This dashboard is waiting for my interpretation.

So, how can we make dashboards do more work for us? How can we make them step into our human world and make our lives easier, rather than forcing us to step into their cold, unfamiliar terrain? Here are some principles to consider…

1. Shift the burden of interpretation

Focus on clearly communicating and highlighting actionable insights about the data. Synthesize underlying data into useful information.

Of utmost importance, clearly label and explain the data view. Explain what metrics mean in layman's terms. Then, as much as possible, shift the burden of interpreting data from the viewer to the dashboard, by intelligently highlighting points of interest.

While automating the generation of deep insights is next to impossible, you can do simple things like highlighting maximums/minimums and recent changes in the current dataset. A smart dashboard allows the analyst to take an appropriate action directly from the key insight.

For example…

November was your most expensive month, with $1,867 of total campaign costs, over budget by $367. (PROMPT: See breakdown of November costs and reduce your spend)

UserX is your most recent conversion to a Platinum membership, 3 hours ago. (PROMPT: Send them a custom greeting and thank you gift)

UserY is your most influential active user this month, with 6,453 Twitter followers. (PROMPT: View UserY’s Tweets about you or mention them on Twitter right now)

You could save $78 per month by upgrading your cloud instance subscription (PROMPT: Upgrade now)

Further, incorporate competitive benchmarks, target metrics, and historical trends into the display where appropriate, so the analyst can get context and easily make judgements on the performance shown in the data.

2. Beauty is nice, but it’s secondary to surfacing valuable information

Don’t go overboard with glossy visual design.

So many pixels, so few answers.

This purple-ish glowy chart might be appropriate for use in a video game… it’s very pretty. But it’s not clear what we’re looking at and why it’s colored the way it is. (Where is zero on the vertical axis? What do the labels along the bottom mean? What’s that hilly area chart in the background? Why are there two horizontal gridlines? Is this sales per day, week, or month? Which elements are controls and which are labels?)

Start with a simple and understated background that’s not competing for visual salience with critical data points that appear in the foreground.

In this first mockup of an imaginary dashboard, all of the data elements have strong, bold fill colors. Color is a very powerful visual channel for directing attention — once you’ve over-applied color, you’ve lost the power of that channel. (Other channels would be things like text, size, position, grouping, and animations like pulsing or moving)

The dark grey social platform labels are almost unnoticeable and the area charts are missing axis and value labels.

In the second mockup of a dashboard showing monthly gas, water, and electricity usage, the display predominantly consists of subdued shades of grey, with sparing use of colored accents to direct attention to the active category (gas) and the latest month (April). It could benefit from the addition of clear labels, but it’s a definite improvement in terms of color.

Both examples by Dribbble user Radium

3. Humanize, humanize, humanize

Provide an uncluttered view with text and wayfinding that’s so easy enough for any person to understand. Clear, punchy language not only simplifies an interface for improved comprehension, but is also an opportunity for achieving an emotional hook that draws the user’s limited attention span.

Just because your dashboard will be used by experts who are familiar with the metrics doesn’t mean you should adopt their opaque jargon and acronyms. The analysts using the dashboard may need to show it to non-experts. Or, they may be new to the job and still learning the subject matter. Or, maybe they are simply tired and grumpy today. Either way, clear and friendly language will kickstart the absorption of information into their brains and make their lives easier.

Avoid overly abstract metrics. Include crystal clear explanations and instructions on how to use the functionality of the dashboard. These can be tucked away in easy to find tabs, modals, popovers, and tooltips.

4. Avoid information overload

Don’t burn their eyes out

Prioritize and arrange information into cognitive chunks, or high-level blocks. Package and fold information into a tree that can then be unfolded as users have a need to investigate further for more details. Allow users to obtain additional information on demand (popovers, tooltips) and drill down as needed.

5. Build views to address specific problems & questions

Base the design on real problems of real people

Listen to a user before you build a dashboard for them.

It’s not uncommon to architect a dashboard — to chose which metrics to include and where — based on purely speculative conversations around a conference room table.

Instead, interview the analysts who will be using the tool. Understand what they do know, what they want to know, and what information you can identify that isn’t on their radar yet. Understand how they currently make decisions and where they get the information required to do so. Anticipate their decision making needs with your design.

As an example, one dashboard built for a company promoting its video game titles on Facebook automatically applied a week-long time window to the data that allowed the user to quickly scroll between weeks. This facilitated the week-over-week comparisons which are common in short-lived social media advertising campaigns.

6. Consider the day-to-day reality of those using the dashboard

Just as a great employee is a person who makes her manager look good, a great dashboard is one which makes its operator look good. The analyst, administrator, or manager who is using a dashboard has challenges and opportunities related to her job. She also has general fears and desires. Think about how you can design the dashboard to make her life easier, to help her wow her boss, to help her get a raise. Think about how you can design the dashboard to reduce stress, uncertainty, and fear.

Think about how you can design the dashboard to reduce stress, uncertainty, and fear.

--

--

P Van
Adventures in Data, Design & Development

Reduce suffering | Teach humans about being human | Donate money