The Problem with Dashboards (And A Solution)

Folks, I’ve been discussing dashboards forever. I’ve consulted with dozens and dozens of clients on dashboards. And through all of that experience, I’ve discovered that dashboards kind of suck.

Don’t get me wrong, I understand the need for a succinct complication of performance on key indicators. It sure beats a 200 page slide deck. But in the race away from the lengthy tomes of the past, I think the pendulum swung too far. These days, the dashboard trend is to try to cram everything important onto a single page. I’ve done it for clients! Here’s one:

AIFSRCDashboard

I have taken many many clients to this place. And they generally feel good about the design. And then sometimes I’ll get the quietly concerned email from the people in my tribe, my fellow research nerds: Um, Stephanie? This looks elegant but we had major data collection issues in Q3 and that’s not obvious in the dashboard and I’m concerned that if the Board sees this they aren’t going understand the caveats.

Friends, these are legitimate concerns.

People are primarily driven by their eyeballs (haha, you thought I was going to say another body part but it really is the eyeballs). People want to look at the graphs and, if the graphs are strong on design and readily understood, people are going to take actions based on the graphs. And this is when the research team freaks out. What about the context? What about how the measurement on this metric changes three times? Yikes, please don’t make decisions off of this quite yet.

But the modern conceptualization of a dashboard leaves little room for those extremely important narrative points of interest. This is the downfall.

As opposed to a single-paged dashboard, let us loosen up to a dashboard report. Awhile ago I edited a journal article written by my friend Veronica Smith, where she outlined the difference. Where a dashboard is a single page of visualized metrics, a dashboard report is a multi-page document with one or two visualized metrics per page and healthy room for narrative. It’s the pendulum swinging back toward the middle, but not nearly so far as those 200 page text heavy doorstop reports. Here’s one page from a dashboard report I produced with some long term clients, the Oregon Health Authority:

Figure10.1_OHADots

You can see we are reporting 2 metrics here and they are accompanied by a text box that holds helpful explanation. It isn’t a burdensome amount to read; it doesn’t even fill up the left hand side of the page. But it gives the research team enough space to say their piece and feel comfortable sending the data off to the decision-makers.

If you are feeling a little too restricted by the current dashboard trend, try a dashboard report. You’ll emphasize the metrics with some high quality visualizations and add a tiny bit of narrative so you can sleep well at night.

12 Comments

  1. Could not agree more with what you say. My question to the BI vendors is why do they all pretty much ignore the need for a simple way for the end users to enter narrative?

    1. Most BI solutions are only as good as the database/warehouse supporting them. DBAs are often terrified of “free text”. That’s why you don’t get free text options. Though you should be able to export the graphics (or refresh) to a dashboard report that supports the narrative.
      The (R)evolution is coming.

  2. This is great! I think too often people confuse dashboards and infographics. Dashboards, as you say, are really about a collection of small visuals like (edata bars and sparklines), and are meant for internal sharing only, with stakeholders who already understand the context. It’s so frustrating when I’m asked to send our current dashboard to some external party, knowing that they won’t be able to make much sense of it.

  3. Priceless!
    Dashboards can be too simplistic and most dashboard graphics suck.
    We’ve been fortunate enough to find a solution that gives us the flexibility to manipulate both the appearance of the graphs and the data that’s included. While that comes with additional setup time, it gives us a workable compromise that I suspect would leave even Tufte and Few with little to criticise. Oh, and we avoid using the gauges and pie charts that all dashboard vendors feel compelled to provide.
    We’ve been producing what you call dashboard reports. We’re trying to strike the middle ground between providing the C-suite with what’s generally wanted in a dashboard (quick information – what’s working, what needs fixing) where others want more detail. We do so in part by adding commentary (still relatively brief and certainly to the point) which has more details as to context and guides interpretation. Adding commentary also allows us to detail the actions we’ve taken to get these results and what’s needed next to improve or fix things.
    But there’s still challenges in dealing with the finite space of a computer screen and keeping commentary succinct, while not trying to add too many extra dashboards to the report.
    This was good to read, thank you — it makes me think we are on the right track.

  4. I appreciate including the measurement description and purpose next to the results. Often, the what and why is separated from decision makers, and even those responsible for data collection. Bringing the explanation to the front gives everyone the foundation for understanding the results and how they relate to the organization’s work.

  5. Hi Stephanie,
    I think the real problem with dashboards is that only a limited set of indicators lend themselves to be represented in a dashboard. Providing a narrative next to a graph is useful, often even necessary to be able to understand the dashboard. However there is information that just can not be displayed through a dashboard.
    I am not a dude though, so I might be wrong.

    Josse
    Dude: The word was first used in the late 1800s as a term of mockery for young men who were overly concerned with keeping up with the latest fashions. It later came to stand for clueless city folk (who go to dude ranches) before it morphed into our all-purpose laid-back label for a guy.

  6. Actually, thinking a bit more about this, this is a lot like Duarte’s “slidedoc” format as a way to bring visuals and short text together for reporting and briefing, except in a dashboard context. Love how the thinking converges sometimes..

  7. Great post, I totally agree. My take on this is similar to the dashboard report idea. But instead of decluttering and extending the classic one pager, I went for separating the content into two sections, starting with an overview followed by the actual data story.
    The starting point should feed and satisfy the users’ curiosity. In many cases they already have some expectations of specific metrics and want to verify them. An interactive and explorable overview section is the quickest (and most fun) way to select and retrieve the information wanted.
    The data story part is supposed to be something like a “scrollable slide deck”, that (ideally) saves people from fighting with PowerPoint.

    Have a look at http://stats.ehpa.org/hp_sales/data_stories/story_sales/
    Apologies for likely bugs and the use of a doughnut chart!

  8. Hi Stephanie – just reading this post about the dashboard report. What program did you use to create it and blend the charts with text? Thanks.

Leave a Reply

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

This site uses Akismet to reduce spam. Learn how your comment data is processed.