Dashboard Design Principles | Beth’s Blog

Dashboard Design Principles

Books, Measurement

Dashboards are a nonprofit’s best friend because they can be powerful tools in communicating your organization’s important measurement data at a glance.   If a dashboard fails to tell you what need in glance, then you’ve wasted all that data collection time.    What if you married with the power of a dashboard with visual design?

Maybe you are scratching your head and thinking, but I’m not a visual designer.     (I’m not either)     The challenge of good dashboard design is that you have squeeze a lot of information into a small amount of real estate and present a display that is easily understood.     That’s not an easy task.

But what if we used design thinking to make our dashboards more effective.     That’s what attracted me to this book Information Dashboard Design by Stephen Few.  It has been around for a while, and it was actually fun to read – as well as having beautiful visual design.

There is a chapter that offers a few principles of good dashboard design that are useful to set up your next dashboard.  These are also good for designing reports.

(1)   No Scrolling

If you can see everything at once, it is most powerful!   Try avoid fragmenting your data by having to scroll.  This requires some discipline to think long and hard about what data is most useful to gain an understanding.    A nonprofit recently shared their social media metrics spreadsheet with me and it was 150 rows of data that require constant scrolling.   If you can see the snapshot and learn, it is useless.

(2)  Give Context

One piece of data out of context on the screen can be meaningless.    KD Paine says that data without a context is trivia.   So think about a series of data that tells the story.    What do you need to compare the data to?   Do you need to a monthly trend?   Can you do it with color coding to indicate whether the number has gone up or down and whether that is a good thing or not.     Supply the right context for key metrics and that will make the difference between numbers that sit  on the screen and those that enlighten and inspire action.

(3)   Don’t Give Too Details

Dashboards should have high level information to support the viewer’s need for a quick overview.   Too many data points or data points that are too precise slows down our processing.

(4)  Choose the Right Measure

Stare at your charts and graphs and numbers.   Is it the right measure.    Always think carefully about the message that most directly support the viewer’s needs and then select the measure that support the message.     Gene Zelazny offers this simple framework:

-Step A:  Determine your message
-Step B:   Identify the comparison.  There are five kinds – component, item, time series, frequency distribution or correlation.
-Step C:  Select the right chart.

(5)  Choose the Right Display

You can show data as a table of numbers or you can present a bar chart, pie chart, or series graph.     Decide which one makes your vivid point.   Another problem is not labeling the charts so you have no idea what is being compared and how.     If you’re looking for some expert guidance in selecting the right chart,   “The Say It With Charts Toolkit” by Gene Zelazny is available on google books.

(6)  Meaningless Color Coding, Variety, or Decorate Elements

I love color coding.  But color coding to for sake of using different colors is meaningless.  Use color codes to help the viewer better understand the data.   Consistency – in this case – is not the hobgoblin of small minds.   Using the same chart or a consistent and meaningful color scheme allows people to use the same perceptual strategy for interpreting the data which saves time, energy, and cognitive overload.

Avoid adding meaningless decoration.  It is distracting and does not add to understanding the data.

(7)   Highlight Important Data

When you look at a dashboard or graph, your eyes should be drawn to the information that is most important.   If you find something important to highlight,  make sure it stands out.   Point to it with an arrow or highlight it in yellow.

What are some ways that you have designed your organization’s social media dashboard to be useful for decision-making at a glance?

9 Responses

  1. Sanford says:

    good article. couple more thoughts:

    1) what gets measured is what gets done.
    2) can’t remember exactly why, but a study said that “bar” graphs don’t provide meaningful information. well..maybe because it’s not contextualized.
    3) keep it simple, one page. be intentional about keeping it simple. ex: if you add a new measurement, then make yourself take an existing one out.
    4) at north point, we always say “what are the three things we want to talk about / measure / discuss / etc”.

  2. Chris Busse says:

    Stephen Few’s book that you mention in your post is excellent, but I’ve had clients not like his approach because his visual style is, in their minds, too sparse. They want “sexy data”. Like many things changing that mindset requires educating the end user (and if you’re lucky, getting them to read the book!)

    This post is definitely something I’ll share with co-workers & clients and I’m linking it up from my (new) social media measurement Tumblr at http://ReturnOnX.com

  3. Josh Dormont says:

    While I am an analytics junkie and work for a non-profit and typically love the information you share, I think you’re approaching data here incorrectly. Your premise is that a dashboard should tell leadership what it ‘needs to know.’ Without defining ‘what it needs to know’, you’ve essentially created a pretty way to collect and report on information. If I am a leader, what I want to know is ‘what can I do, what are your recommendations?” not “oh, looks like we had an increase of 15% new visitors to our
    site.”

    I would challenge you to use that dashboard to answer the question of “what decisions do I need to make?”. As analysts (or not) we should focus on helping people make data-driven decisions. Does the dashboard do that?

  4. Ryan says:

    Good summary Beth. New tools and applications are emerging and the report as a dashboard will soon be replaced with a database enabled applications. Design is being standardized. A great example from an application being designed for HR professionals can be found at http://www.visiercorp.com/
    The premise is to plug in with multiple data sources that are integrated through one application.
    This is an exciting time with evolving balanced score cards, improved design and enabled by databases that are accessible through the web. What happens to organizational leadership when knowledge is democratized? I presume good things!

  5. Beth says:

    Josh: Your point is right on!! Thank you for helping to clarify that! Top design decision.

  6. Noah Flower says:

    This is a great start on a set of general principles. But I heartily agree with the comments about the importance of giving careful thought to how this data is going to be used for decisionmaking. What you can measure is often quite different than what you need to know in order to make a good decision. I think a dashboard can only really be useful as the extension of aligning the management around their key priorities and ongoing choices. And I think that it should not just be about the data, but start with the decisions and show the hard data as one input alongside questions that can only be answered qualitatively. In reality, dashboards are most useful not for strategic questions but for tactical questions, where there are more opportunities to have data drive decision-making directly.

    But to the question of how it looks: I fully support the war on chartjunk. Edward Tufte and Stephen Few have it right — get rid of all the excess visuals and make the visuals convey the information they contain as clearly and directly as possible. Tell the client that they’ll learn to love it! The more you dress up the data, the more people will treat it like a literal dashboard and fetishize it as the only thing they need to know.

  7. Stephen Lynch says:

    I would challenge the idea that a dashboard should attempt to answer “what decisions do I need to make?”. A one-screen display can answer the question “What changes, expected or unexpected, are occurring.” It might just be able to hint at the answer to “Why is this change occuring?” If a one-screen display can answer the question “What action should I take?”, then you’re running very simple organization or system.

    Expecting a dashboard to have some sort Artificial Intelligence is going to result in failed projects and unfulfilled promises.

  8. Calvin Wells says:

    I agree that dashboards can be very useful for helping organizations to visualize and react to key performance indicators. This type of data visualization occurs at various levels and departments of an organization, including Information Technology which is where I’ve leveraged dashboards most often. In response to Stephens comments, I think a form of dashboards with integrated decision algorithms are already heavily used in the financial markets. For other markets, if the objectives are precisely defined in a scorecard fashion, I don’t see why a decision algorithm and the accompanying dashboard can’t be integrated. I don’t think it’s a matter of artificial intelligence where it has to be a system of reasoning. I think the algorithm that drives the dashboard is derived of reasoning and logic that matches the dynamics of the interest.

  9. Saju C Mannarath says:

    Wonderful discussions above for the crisp article. Dashboards should help the ‘Drivers’ to keep moving steady and safe. It can also indicate the extend to which the vehicle (system) can be put to test or accelerated. Hence, the Dashboard is not above the engine or the maker of the vehicle. it is only an indicator which is useful for the driver, as well as to the manufacturer to some extend, to test more options (speed,efficiency, etc). If the drivers role is to check the veracity of the vehicle he is doing a research for his Boss. If he is doing a safe driving, he is keeping the promise his Boss has already given… So, the Dashboards can be used differently; for research or for monitoring smooth performance.