Forty-eight people illustrated alternatives to a secondary y-axis in November as part of the first #SWDchallenge taking place in our exciting (and still-in-beta) SWD community! The team and I are thrilled, both with the level of participation in this new venue and in particular about the robust dialogue it makes possible. Now that community members can provide context on their own submissions and add commentary to others’ creations, we’re having some great conversations and seeing productive iterations.

One of my personal hopes with the community is that it be a space where people can do exactly this—share their work and get thoughtful input—and that we can all learn from this process. It was amazing to see that starting to happen with the challenge. Big thanks to everyone who contributed their work and provided feedback to others. I look forward to the community conversation continuing to grow over time.

I promised signed copies of Let’s Practice! to five lucky (randomly chosen) participants. Please join me in saying congratulations to book recipients Brent, Armani, Dennis, Eddie, and Amanda (folks: stay tuned for outreach from us to get your shipping info)!

We enjoyed reviewing all of the November submissions. People generally employed one of the two alternatives I put forth in the challenge description: (1) preserving the secondary y-axis but not showing it and labeling data directly or (2) separating into multiple graphs. There were also a few other approaches employed. Alexander and Ben used color-coding to make it clear which data goes with which axis. Julia found a clever way to overlay lines on top of bars and leverage a single common y-axis. Simon and Pris each used sparklines together with bars, to give a sense of the trend and allow for easy point-in-time comparison across categories; Adam employed a twist on this, using a bump chart together with bars.

Speaking of twists, there were also some interesting ones on norms—for example, Augusto contributed a view where time on the vertical axis makes perfect sense and Charles switched the typical lines-then-bars design to provide a wealth of data on worldwide sunshine.

I appreciated seeing many real-world work examples (and work-inspired examples). Claire submitted one of her iterations to improve a visual from a colleague’s report; Jack contributed a viz that reportedly changed his career. Jeff showed an approach for providing additional context around summarized survey results and Peter put forth a number of potential views for headcount inflow, outflow, and volume over time. Klaudia and Stela each visualized website performance, clearly annotating important events and context. Laron welcomed feedback on visuals used at his organization for assessing risk associated with tourism, while Ishan shared a customer dashboard he developed. On the personal front—and a great example of using data to understand something better and drive behavioral change—Joost illustrated his own screentime data, forming new insights about his habits. There were also many good examples that employed takeaway titles and used color effectively to differentiate data series and visually link annotations to data.

A number of people were creative and had some fun in ways inspired by their chosen topic (including a variety of food-related visualizations, perhaps resulting from coinciding with the US food-focused holiday). Armani employed a honey style to show data related to production of the sticky stuff. Kate M. dove into a meaty topic (complete with fab drawings of steak!) and shared some data-related challenges she encountered along the way. Lori chose pink and waffled fitting for her ice-cream theme (great use of icons, too). Kate S., Kolyu and Robyn each incorporated relevant images in their visuals, while Paulo embellished his bars with subtle gas pump details.

On the cautionary front, there were a couple practices I observed that warrant mention. One challenge that arose for several people who opted to preserve the secondary axis but hide it and label directly is overlapping data series. This can inadvertently draw attention to parts of the graph (for example, where bars rise above a line) that are simply the result of where the respective minimum and maximums across the two y-axes are set. There’s not always an easy fix to this, but it’s a tradeoff to be aware of when encountering this scenario and deciding to pursue this approach in light of it. I also noted a couple of instances where it looks like smoothing was applied to line graphs. While this can make the data look kind of cool, it’s usually not good idea. There are some cases where applying a smoothing function makes sense from a statistical standpoint (for example, if seasonal adjustment is warranted), but smoothing for the sake of creating a nice-looking line is not advisable. Doing so takes data that accurately measures something (for example, a straight line that illustrates the relative change from one month to the next) to something that no longer depicts reality.

Shifting back to the positive, you can see all 48 submissions in the SWD community (if you haven’t already joined, simply indicate interest in the form that appears and we’ll follow up with information on how to log in). As you browse entries, I highly encourage you to leave comments and add datapoints to work you enjoy.

Again, big thanks to everyone who took the time to create and share their work and those who lent their feedback to others. Be sure to check out the December challenge, underway now!