Monday, March 19, 2012

lessons from GMN

GMN is the Grant Managers' Network. This afternoon, I had the distinct pleasure of speaking at their annual conference in San Antonio.

My topic was storytelling with data, presented in two 90-minute sessions to about 200 super engaged grant managers. The first session was an overview of data visualization best practices. The second was a hands-on workshop focused on applying the lessons covered in the first session to real examples (submitted by participants) to practice employing those freshly learned skills for showing philanthropic impact through storytelling with data.

With this post, I thought I'd recap the lessons I covered today and show them applied to one of the visual makeovers that we focused on during the workshop session.

First, the lessons:

  1. Choose the right type of display: leverage bar charts whenever possible due to ease of interpretation, lines are for continuous data only, avoid pies
  2. Eliminate the clutter: get rid of the stuff that doesn't need to be there, de-emphasize the necessary but non-message impacting stuff
  3. Focus attention where you want it: leverage pre-attentive attributes (color, size, thickness, enclosure, placement on page) to draw your audience's attention to the important parts and create a visual hierarchy of information
  4. Think like a designer: include affordances that make it clear to your audience how they should interact with your visual, make the visual accessible by favoring simple over complex, take the time to make your visual aesthetically pleasing to gain your audience's attention and patience
  5. Tell a story: don't use graphs to show data, use graphs to reinforce your story; make your story explicit with words

Now, let's take a look at these lessons applied to one of the participant-submitted visual. Here's the visual:

There is a lot of information here. Here's a glimpse into my thought process as I look at it and start to figure out how I want to approach turning it into a story:
  • I'm unsure at first how the data in the graph and two tables are related. The first thing I did was some math to try to better understand how the numbers relate to each other (to learn/verify things, for example Adoptions + Transfers + RTO = Live Release).
  • Upon closer inspection, it seems the overall story is around positive (live release) and negative (euthanasia) outcomes; I want to make these two sides of the story more immediately visually clear.
  • There are some unfamiliar things that I want to clarify: the acronym RTO and the fiscal year dates (so I know whether/how to compare Jan to FYTD). A quick email exchange with the submitter later, I learned that the former is "Released to Owner" and the latter is Oct-Sep.
  • Depending ont he story to be told, there may be too many comparisons. The way it's set up now makes me want to compare Jan figures with fiscal YTD figures, which probably doesn't make sense.
  • I crave an action title: the top-of-page space is precious because it's the first thing an audience encounters. I should be used to tell the audience what they need to know and orient them to the information that will follow.

After sketching out some things on paper and iterating a few times in Excel, here is my makeover of the visual:


Let's take a brief look at the changes I made, according to the 5 lessons I outlined at the beginning of this post:
  1. Choose the right type of display: This isn't necessarily the right display, but it's one approach. I combined the data into a single visual and made life and death visually opposing (note that death is both red and in the negative direction to reinforce visually that it is a poor outcome).
  2. Eliminate the clutter: I reduced the visual comparisons by focusing on FYTD and eliminating the January comparison. I pushed the axis and axis labels to the background by making them small and grey so they wouldn't compete for attention with the more important parts of my visual.
  3. Focus attention where you want it: I made use of pre-attentive attributes: bold, color, space, and size of text to draw the audience's attention to the important parts.
  4. Think like a designer: In this case, this was mostly about paying attention to detail: making sure things were aligned, leveraging white space, and trying to make the visual as easy to understand as possible.
  5. Tell a story: I added a story in words to make the message clear.

I also did another version with a very slight twist to show how relatively small changes can completely reframe the overall message:


What do you think? If you attended one of the sessions, I welcome feedback on what worked well and what could have been better. (Other comments welcome, too.)

Stay tuned - I'll post about the other makeovers in the coming weeks. If you're a first time visitor to my blog, you can sign up for email updates in the upper left. I'll also point you to a couple popular posts: how to do it in excel and no more excuses for bad simple charts: here's a template. Happy storytelling with data!

12/5/13 update: To download the spreadsheet with the visual above, click here.

7 comments:

  1. I discovered your blog a few weeks ago while Googling your Google team. Thanks for sharing all of your advice!! I'm already making prettier Excel charts, and I look forward to reading each of your posts.

    With the charts above, one thing that is not intuitive to me is why the lines on the left side of the chart (death data) are thicker than the lines on the right (life). Quickly glancing at the chart, I would assume that the data on the left side should be weighted more heavily or are more important than the data on the right side. What are your thoughts on making the lines on the left side thicker? I think the graph might make more sense if both sides used lines w/ the same thickness.

    There's some research that shows that people automatically infer a lot about social power based on the length of lines that are used to illustrate data (if you're interested, Giessner & Schubert, 2007 have some cool research on this idea). Although I think most of this research looks at the length of lines, I would not be surprised if the width of lines were equally important for social inferences.

    ReplyDelete
  2. I would have to imagine that Cole hadn't intended them to be different thicknesses?

    ReplyDelete
    Replies
    1. Your assumption is correct! That was an oversight - I had thickened the bars on the left and somehow forgot to do the same to those on the right. Oops. (No one is perfect, right?) :-)

      Delete
  3. Cole - your session was awesome. I only wish we would've had onsite computers to practice in part 2! Thanks for recapping everything so thoroughly. Can't wait to take these lessons back to my team.

    ReplyDelete
  4. Everyone who attended your session was very ashamed of their graphs. Tuesday's presentations were filled with "now, I attended the data visualization session, but didn't have time to redo my graphs."

    clearly made an impression!

    ReplyDelete
  5. Can you illustrate how you set up the data in tables to achieve the side-by-side effects?

    ReplyDelete
  6. Thanks for your comment! The final visual is actually two horizontal graphs shown side by side (with death plotted in the negative direction and life plotted in the positive direction) and formatted to look like a single visual. To get this from Excel into another application (e.g. PowerPoint), I copy the entire block of cells and paste into the ultimate application as an image.

    I've added a note to the bottom of the original post with a link where you can download the Excel file to poke around and see how it is set up. I hope this helps!

    ReplyDelete