(I know some of you are cringing as your brain conjures up examples of those sloppy, overdone, chaotic messes with five different fonts.)

As calendars start to fill, I wanted to flag three early-ish September events that will help you communicate data more effectively.

Subscribe now

Data Viz for Public Health Sprint Workshop

Bill Bicknell, one of my most memorable professors at BUSPH, opened his global health policy class with a slide that read:

His point was that the decisions we make with data shape who has the knowledge to take care of their own health, the policies that make health care more accessible, and the access to care that allows people to connect to services in equitable ways.

Data visualization plays a big role informing those decisions.

If you’re a public health data nerd, this 2 hour workshop is for you. It’s a masterclass in design lessons that are immediately applicable across different audiences and use cases for visualizing data, all with examples from the world of public health.

Plus we’ll talk about the real world challenges of communicating across audiences, with different goals, and when lives are at stake.

Friday, September 18 from 11AM-1PM EDT, hosted virtually on Maven so you can join from anywhere. Register with the code ‘vizresponsibly’ for a special discount for newsletter subscribers.

Chart Chat 74: From Research to Practice

I love research papers. Probably goes back to my time working in monitoring, evaluation, research, and learning roles in public health. But I know that’s not the case for a lot of colleagues who have spent their careers in the private sector or government.

During this month’s Chart Chat, Andy, Jeff, and I will share impactful papers that have shaped our design decisions and how we teach data visualization. While we’ve hosted some incredible researchers like Stephanie Woerner, Steve Franconeri, and Eli Holder in the past, the world of data viz research is vast.

Thursday, September 10 at 11AM online. Register on eventbrite to get a calendar invite and reminders so you don’t miss the live chat, or find us on YouTube or LinkedIn.

“Make with AI. Make Mistakes with AI.” on the next installment of The AI Analyst

Running in reverse chronological order Andy is hosting data viz expert and former DVS Executive Director Elijah Meeks on the next installment of The AI Analyst.

Elijah and I collaborated for years on the DVS board and he’s a data viz expert not afraid to be provocative in his thinking (and LinkedIn posts) about the ways AI is reshaping parts of our field.

Their conversation will focus on making stuff and making mistakes with AI tools, which is arguably one of the best ways to learn the strengths and limitations of what’s out there today.

Thursday, September 3 at 11AM EDT, online. Register on eventbrite for a calendar invite and a reminder about the livestream.

A note about projects and working with me

This year I’ve worked with a number of teams across education, public health, foundations, and pharma on different analytics, data visualization, and design projects.

While I still do some hands-on design work, these days more of my projects focus on advising teams around a sticky data use challenge. Sometimes that’s redesigning and entire digital reporting ecosystem and coaching on user-centered, Agile approaches to build dashboards someone actually wants to use and other times it’s reimaging reports into snackable data communication products.

I also continue to offer private workshops, training sessions, and public speaking on visualizing data responsibly.

I’m currently booking projects into 2027. If you have a sticky challenge around how teams communicate and use data and are looking to bring in an external perspective, I’d love to chat. Drop me an email, and I’ll send over my booking link to find a time.

Subscribe now

This week, let’s talk about election charts that I’ve been spending way too much time browsing as a Wisconsin voter and data nerd.

Subscribe now

About the race

The Wisconsin democratic primary for governor is coming up on August 11. One of four candidates still in the race will go on to face Republican Tom Tiffany in the general election in November.

But the last few weeks have been a wild ride if you’re a Wisconsin voter. We’ve had two major polls in July from Marquette and State Navigate/Main Street Alliance/Up North News. Both happened before the debate on July 28 and before additional departures from the race.

Some of that data has been buried in long tables, but there have also been some interesting charts that address data communication challenges common in elections and other survey data:

• Undecideds have remained a large response group, resulting in seemingly simple but highly impactful choices around where to rank the undecideds on bar charts.

• Second choice preferences are particularly important when candidates drop out in the middle of or after a poll, creating one of the best uses of a Sankey chart I’ve seen outside of finance and budget data.

• Sentiment reporting, including favorability on issues and candidates.

In theory, everyone should vote for the candidate they want most. That’s the real point of having a primary.

But charts have the power to sway perceptions around electability and issues. As you read through reporting in this mid-term cycle, consider how you’re using the data as a voter, which may be different than how campaigns use the data to shape messaging and fundraising.

The changing field

I’ll get to the charts in a moment. It’s helpful to have a bit of context on the timeline of events over in the last month if you’re outside of Wisconsin and haven’t been following the shakeups.

A month ago, we had six candidates still in the race. Then…

• David Crowley (Milwaukee County Supervisor) dropped out and endorsed Sara Rodriguez (current Lt. Governor and winner of the straw poll at the WisDems convention). (WPR)

• News surfaced that Rodriguez’s campaign had some financial inconsistencies with less cash on hand than reported.

• Rodriguez drops out.

• Crowley re-enters the race.

• The only official debate happens with five candidates on stage.

• Former Lt. Governor Mandela Barnes drops out, with no official endorsement of any of the remaining candidates.

There’s a crew of four political creators who have been doing amazing ‘group chat, but on YouTube’ discussions about the state of the race if you want to dig into the details of the last few weeks.

So a week from the election, we have four candidates still in the hunt, with Francesca Hong (WI State Assembly member) far in the lead in the polls and gaining national attention with coverage in the NYT, Washington Post, and more as a democratic socialist leading the race for governor in a swingy state.

The other remaining candidates include Kelda Roys (WI State Senator), Joel Brennan (former WI Secretary of Administration), and Crowley.

With that context, let’s unpack a few notable charts and design decisions.

#1: Reporting on undecided voters

Over the last month, undecideds have remained a large response group. So what’s the best way to represent them in poll reporting?

Remember: sort order often shapes the story on bar charts.

Many poll charts have followed the common convention of sorting an ‘other’ or ‘don’t know’ category as the last bar in the chart (below, left). This approach focuses our attention on who is leading among the candidates.

But when nearly half of voters remain undecided, that group may be the story, as we saw in the results from the mid-July Marquette poll.

To focus attention on the large share of undecided voters, moving that category to the top of the chart focuses attention on the potential for change. Adding a clear headline and highlighting with color focuses attention even more (below, right).

When the share of undecided voters is larger than multiple candidates, my personal preference is including undecided ranked in the mix (right) since they’re a large and important block.

#2: Plotting second choices

When we’re in a general election, candidates don’t change often in the window leading up to Election Day. The two major parties pick their candidates, and you may have a few other options on the ballot.

But primaries are bigger, more crowded fields with candidates exiting on an ongoing basis. Adding to the complexity, candidates may drop out when early voting has already started. As a result, some voters may have cast ballots for non-viable candidate, which is one of the best cases I’ve seen for rank choice voting.

Peek back at the bar charts above of the Marquette data: the second (Barnes) and third (Rodriguez) ranked candidates are no longer in the race. (Plus, another candidate—Crowley—returned to the pack who wasn’t included in the poll.) So where are supporters likely to shift for Barnes and Rodriguez?

The more recent poll from Up North News, Main Street Alliance, and State Navigate dedicated time in the reporting to those second choices.

In their poll, conducted July 23-26 among 1,085 likely voters, Francesca Hong led (44% of likely voters). But the data on second choices for Barnes supporters become very relevant when he ended his campaign. A quick look at that support looks like another boost for Hong, which is what comes out in the recalculation of support.

The precision of the bar chart is really helpful for looking at who stands to gain the most from Barnes’ departure, but State Navigate’s Sankey chart is a standout among the poll charts for me.

You can see the flow between candidates, it’s visually engaging, grabs your attention, and rolls up the totals for potential additional support. And the clear ranking on both the left and right based on share of the vote makes it easy to quickly look for who’s leading now and who has the biggest potential gain from other candidates’ exits.

What the Sankey doesn’t do well is show the detailed distribution of where supporters go when a single candidate drops out. Great chart for engagement, less for precision.

The whole report is really well done, particularly their consistent use of color for each candidate so you can follow one candidate across charts. Take a peek!

#3: Reporting on issues that impact electability

Polls about candidates aren’t the only ones that shape opinions and behaviors of voters though. Issues matter too.

The July Marquette poll included some issue-related questions alongside respondents’ candidate preferences. With Hong leading in the polls, the big one that grabbed headlines asked about perceived favorability of democratic socialism and the DSA as a party.

When we ask for favorable, not favorable, and undecided responses about an issue, my go-to is a diverging bar chart. Diverging bar charts are great for showing shifts in sentiment across categories, and have an implicit ‘negative / positive’ direction that relates to our mental model for sentiment.

One drawback is that diverging bars don’t put the two categories on an aligned scale though, so estimating the difference between favorable and unfavorable is a bit more work than on a dot plot.

But back to the theme of undecideds: where do we put the share who need more information? Steve Wexler, who is one of the industry experts on survey data design, has written about the benefits of splitting the neutral category out separately, as seen in the charts below and and it’s an approach I use often.

Who is represented in the data is important too. Across the Marquette poll, the results includes a mix of data for registered voters and for Democratic voters.

If want to assess what the overall favorability of an issues is across the full voter base, including independents, looking at all registered voters could be useful when considering electability for Hong if she wins the primary, as in the chart above.

But there’s probably a reasonable share of people who hold an unfavorable view of democratic socialism AND are not going to vote for the Democrat no matter who the candidate is. So perhaps we look at democratic voters only, including considering their political leanings.

The additional data I wish I had to make sense of this chart is how large each of these subgroups are, but a girl can only dream.

So why do charts on issue favorability matter? Will people change their opinion about something like democratic socialism just because they see others like them have a positive opinion?

Maybe.

Eli Holder, an exceptional data viz researcher and communicator, has explored the effects of seeing charts of opinions aligned to different political ideologies. In a recent study, key findings from research on public opinion polling, the two big findings were:

Functionally, that could mean that seeing overall favorable opinions of democratic socialism could pave a clearer path for Hong. Or, that seeing unfavorable opinions could make the road tougher, particularly when trying to persuade conservative-leaning or independent voters.

I have no crystal ball though, and a lot can change day-by-day on a campaign. I’m here to talk about charts, not make political predictions.

If you’re a data nerd or even vaguely curious about how visualization research is conducted, Eli is one of the best in the field at creating plain-language explainers about his work, including the details behind the opinion polling research.

The lesson: be mindful with poll data.

Discussions around ‘is it even worth voting for my candidate’ keep surfacing as Hong pulls an increasingly wide lead in the two July polls.

My opinion? Yes.

All of the static charts we see in the news and on social feeds can create a sense of certainty in the polling data. But polls don’t vote, people do.

Want to make the case to yourself that polls can only tell us so much?

The analyst at Badger Brief has put together an interactive tool to play with the WI primary polling data, where you can decide what goes into the model. Dangerous if you’re trying to make predictions, but there’s a very specific disclaimer: “This is not a predication. It combines public polls into one picture and shows how much is still uncertain…The movement is the point.”

Interactive tools shine with ways to emphasize how different weightings of inputs reshape projections, more for entertainment than anything else. But the section of the forecasting tool I appreciate the most are the caveats at the end: “Is this legit?”

That’s visualizing data responsibly in my book: giving the context and framing someone needs to interpret the data, which is just as important as the chart design.

The case for experimenting with AI tools and integrating them into our workflows seems to center on:

We can make things faster (and maybe better?). This can be a good thing, but only if we don’t consider speed or efficiency as the only measure of success that matters. Data viz instructors who I admire deeply have been sharing about their experiments with genAI tools for brainstorming and rapid prototyping to some amazing ends.

The act of analysis and visualization is more democratized. You don’t need to know how to code to vibe code apps with prompts, and you can spin up a chart or infographic without actually touching a data table or tool. Elijah Meeks has been making the case that this is a particularly good thing that opens up the world of more creative visualizations (i.e. the stuff that’s not in the default charts in Excel-ish tools) that may be more engaging and effective.

We need to, otherwise we’ll get left behind. (My least favorite.) Yes, there’s a lot ot explore and learn, and it can feel like we’re getting ‘left behind’ when we talk to folks in tech and some sectors. But talk to people outside of tech like I have at our PTO meetings or over beers, and there are plenty of folks who don’t immediately know what I mean when I say Claude.

Subscribe now

A framework for using the new tools responsibly

Experimenting with new tools requires some guardrails though, particularly when those tools can quickly produce beautiful outputs that seems oh so believable on the surface.

On our latest episode of Chart Chat, “Supercharge or Sabotage?”, I shared my framework with nine considerations for anyone using AI tools in your analytics workflow. They fall into three buckets:

• Audience, to think more deeply about who you’re reaching and their relationship with the data you’re sharing

• Scale, to think about how widespread the damage could be if we shared inaccurate or misleading information

• Sources, to think about your relationship with the data, how much you trust it, and your ability to spot errors.

Rather than being an exhaustive list of considerations, take these as a starting place for discussion, particularly if you’re working on a team.

Each component is placed on a relative scale where the precise position of your rating matters less than the overall picture: dots lean more left, there’s less risk, while dots leaning more right means more risk.

Here’s a quick breakdown of the matrix.

Audience

Ask yourself:

• Who is the audience? What is their relationship with the data you’re presenting?

• What will the data be used to do?

• Who else might be impacted beyond your intended audience?

Scale

Ask yourself:

• What kind of actions are people taking with the data? Is the output more for entertainment or sharing fun stats, or informing real decisions with big consequences?

• How wide reaching would any harm be? How long lasting would that harm be?

• Could harm be remedied in the future or would it be hard to course correct? Here, you need to consider how your work is fitting into the bigger picture of information sources for your audience.

Sources

Ask yourself:

• How well do you know the data? Where is the data coming from?

• Could you spot a sneaky error?

• What bias could be baked into the data? Could you control or adjust to address any bias?

This third category is unique, because you can manage the overall risk of harm due to the audience and scale considerations with subject matter expertise, knowledge about the data sources, and minimizing bias.

What to do when a project is ‘higher risk’ for harm

High risk doesn’t mean don’t publish. Many important, impactful visualizations would score as higher risk for harm, from climate visualizations to outbreak data to key data points for multimillion dollar acquisitions within companies.

Higher risk means take more time and care to minimize those risks and ensure the data you’re putting out into the world is accurate and communicated effectively. Take added care with quality assurance checks, assessing bias in the data sources, and gathering feedback from different people who may have questions about your charts that the team didn’t anticipate.

If a project has a high risk for harm and is just an experiment for your own learning though? Pause before you publish. Consider if there’s a smaller community space, like the Data Viz Society’s critique channel over on Slack, where you can share and ask for feedback to narrow the audience and minimize risk of harm.

Risk frameworks in a world where technical barriers to design are disappearing

Reading through these considerations as readers of a newsletter called Viz Responsibly you may ask the obvious question: aren’t these just good considerations for any data project?

Yes.

So why write about risk in the context of using AI tools then?

GenAI tools are removing technical barriers to create charts and answer questions with data. Charts that required complex code previously can be spun up with a prompt. And they often look beautiful! Even when they’re inaccurate, which is a conversation about beauty and truth for another post.

But just because you can make something doesn’t mean you should publish it if there’s a risk of doing harm through your work.

Tools are also good at executing on prompts and applying design rules (sometimes…), but questions around the risk of doing harm are more nuanced and complex, requiring a human touch.

As more people experiment with data analysis and visualization within these new tools, the risk of harm grows. Humans make mistakes, and so do AI tools. This isn’t an accusation of intentional malice (though the barriers to designing purposefully misleading charts is lower too), but because these new tools are fun! Exciting! And how could a vibe coded dashboard cause harm anyways?

That question sends me back to the peak of the COVID pandemic, when misinformation was creating challenges around getting people to adopt prevention behaviors including vaccines.

Crystal Lee and her team at MIT did a study looking at how anti-mask sentiment moved through networks, including the role of charts. One of the findings was around trust in the chart creators, with some communities banning ‘mainstream media charts’ in favor of self-made graphics. They trusted themselves to make more accurate charts than experts.

Surprising? Not really. The messenger (in this case, the chart creator) matters. But that doesn’t mean everyone who could build a chart could build an accurate one.

A call for more data coaches

As the community of people creating charts grows, data professionals have a responsibility to be better coaches.

Perhaps we do think about all of these risk considerations as part of our workflows for new projects. But how can we make that thinking more visible and accessible to folks experimenting with chart design?

In Data to the People’s 2025 Global Data Literacy Benchmark, they found only 13% of respondents could coach others on data literacy topics, leaving us with ‘too few leaders to support too many learners’. As Jane Crofts (Founder and CEO of Data to the People) captured in her opening letter for the report:

AI readiness depends as much on human competencies as on technical infrastructure.

The integration of AI into day-to-day work, from co-pilots and process automation to advanced analytics and algorithmic decision-making, requires not just technological deployment but human discernment. Employees must be able to read, question, and validate data-driven insights. They must know how to apply context, exercise judgment, and communicate conclusions. And, increasingly, they must be able to oversee the machine.

If we do not act now, organizations may face challenges when deploying AI without adequate oversight, embedding bias instead of equity, and reinforcing decisions that lack clarity or context.

Human capability is not optional; it is a critical safeguard in an algorithmic world.

Here’s a consolidated version of the risk framework, in case you want to try using it when you’re working on a new project. I’d love your feedback in the comments or over email, as it’s a working concept.

Remember, it’s the conversations around these questions that matter even more than any precise mapping on the continuums.

And here are two examples of a completed matrix, for a quantified self project (left, lower risk of doing harm) and a project communicating outbreak data (right, higher risk of doing harm, but with the capability to mitigate some risk because of subject matter expertise). We talk through these in more detail in our Chart Chat episode.

While the first two weeks of June have been all about wrapping up the school year (solidarity to other parents!), the rest of the month is packed with data design goodness. Here are the three events where you can find me in June!

Data Visualization for Public Health Public Workshop

When: Friday, June 12 from 11AM-1:30PM EDT

Where: Virtual. Learn more and register on Maven.

Who should join: Anyone working in public health, education, or non-profits who creates visualizations and knows they could be better. You’ll leave with a clear framework you can apply to every chart you create.

About: A short sprint workshop hosted on Maven with a focus on the fundamentals of good data communication. You’ll leave with tricks to design better charts reinforced with quick hands on activities, all packed into 2.5 hours.

I’ve often heard from public health colleagues that paying $500+ for a workshop just isn’t in the budget, but many more affordable courses and workshops are all asynch and they really want to engage with an expert instructor on the topic for their sticky questions. This workshop is my effort to fill that gap: use code ‘vizresponsibly’ for 50% off the public price, which reduces the cost to under $150.

Plus, all students will get the option to book 30min of 1:1 time with me in June or July to help put the workshop principles in to practice.

Register for the workshop

Chart Chat 71: Supercharge v. Sabotage

When: Thursday, June 18 at 11AM EDT

Where: Virtual - sign up on Eventbrite for reminders, stream on Youtube or LinkedIn.

Who should join: Anyone curious about genAI and agentic analytics use in practice. You’ll get four different perspectives on the benefits and risks of incorporating genAI tools into your data work, without any sales pitch for some new tool.

About: The experiments are everywhere: which LLM does the best with data analysis? How are genAI tools changing how we create dashboards?

Join Chart Chat LIVE for a conversation with four data experts who go beyond showing AI chart examples and debate the bigger picture around whether genAI tools are supercharging our capabilities, sabotaging analysts with sneaky errors...or something in between.

After the show, attendees can pop into our unrecorded after party for an off-the-record chat sharing your thoughts on the supercharge v. sabotage debate.

Register for Chart Chat

Outlier Conference hosted by the Data Visualization Society

When: June 24-26, 2026, with an optional pre-conference challenge hosted in collaboration with the UNDP on June 23!

Where: Virtual. While the last three years have been hybrid events, keeping us all virtual this year means the team has doubled down on creating the best virtual conference experience ever, with plenty of activities and networking beyond the amazing mainstage speakers.

Who should join: Anyone curious about data visualization, from business analysts to artists. The agenda has something for everyone, and is single stream, meaning no compromising on which talk to pick across a long list.

About: Data visualization is never finished but evolving. This year’s Outlier conference, The Final Draft,

is about iteration, progress and the transformative power of curiosity, feedback, tools, context, and constraints. Join a global community of data viz enthusiasts as we learn from real-stories, hands-on experiments, and forward-thinking practices. Whether you are just starting your journey or shaping the next frontier, Outlier 2026 is your space to connect, learn, and share how data becomes visual.

Learn more and register

In today’s world, I think connecting in real life (including in virtual spaces!) matters more than ever. Hope to see you at the workshop, Chart Chat, and/or Outlier this month!

But a few issues showed up on a genAI-enabled redesign that have me thinking about the value of sketchy, imprecise prototypes and the sneaky mistakes that can be missed when we can prompt our way to a beautiful dashboard quickly.

As using AI tools for chart design and dashboard development grows, we have to remember that prototypes aren’t the same as products, even when they look really good.

Subscribe now

Shifting landscape of tools

In past years, these makeover episodes usually involve a lot of Tableau. Jeff and Steve are Tableau Visionary Hall of Famers, Andy was an Evangelist there for nearly 15 years, and they can make magic happen in Tableau. I’m usually the odd ball mix of Excel, Canva, Figma, and, yes, sometimes Tableau.

But for episode 70, all three of my co-hosts used Claude to execute their visions for the redesign. We all used a riff on a line chart; Steve and Jeff built out entire dashboards, Andy crafted a slope chart, and I created a simple line chart in Excel (maybe being a bit contrarian…but it’s still one of the most widely used analytical tools in the world!).

My co-hosts each prompted their way to specific visions for a remake using Claude and created impressive prototypes in far less time than it would have taken in Tableau, by their own admission. The prototypes were impressive not because the tools produce perfect data visualizations, but because of the thoughtful prompting, rooted in their expertise.

I’ll add an aside here: it’s very much worth watching the full episode to see the process breakdown, especially if you’re new to experimenting with genAI tools for building analytical applications. You’ll see the logical sequence of iterations and prompts from different designers.

Prototype or product?

I use the word prototype deliberately.

A prototype is a representation of an idea that we can use to get feedback and shorten our feedback loops with users.

Rapid prototyping, where we spin up messy representations quickly, lets us fail fast and iterate without heavy investments of time and resources. Speed matters, precision is less important (you’re not working with real data). genAI tools can be really useful for creating those interactive visual representations of an idea, getting visuals in front of users for feedback sooner.

In our Chart Chat after party (an unrecorded hang after the show), we talked about hallucination issues with LLMs and potential issues with incorrect calculations. As wrote about on How to Speak Data, with a closer look there were multiple errors that we hadn’t even noticed on the live stream on one of the Claude dashboards.

Which left me musing on a bigger question: what happens when the prototype created in 30 minutes looks so polished and most of the numbers check out?

Creating prototypes on white boards, paper, or in digital tools is a powerful part of the design process.

In my own practice, nearly every dashboard I make with a client starts with mockups and prototypes sketched out and then built in Figma. At this stage, things are wholly conceptual or I’m working with samples from the datasets or dummy data. Each iteration gives us early user feedback on the design, and sometimes surface data requirements that change how we model the data. The earliest versions (which definitely take more than 30 minutes to sketch out in most cases) aren’t pixel perfect.

Now, we have new genAI tools in the mix which can produce beautiful prototypes quickly. But there’s a difference between prototypes we spin up for quick feedback and a final, production ready product populated with real data. We know that as designers, but that may not be apparent at a glance.

In the 30 minutes it would take to sketch out a few iterations of a design, Claude produces an output that looks polished and ‘final’. It looks ready for prime time (or at least ready to share on Chart Chat). And wouldn’t we be offsetting the efficiency gains if we have to comb through and check every number?

What does this ‘polish’ on a rapid prototype mean for analysts and designers?

Experiments with genAI often focus on speed and efficiency as big value propositions. With more pressure to create faster, we move further away from the underlying data and outsourcing the learning that can happen through the creation process. What happens when a prompt can generate a beautiful looking prototype in minutes?

There’s an unfinished messiness in our hand-sketched mockups that’s missing when we prompt our way to a dashboard in Claude.

We also spin up digital prototypes in Tableau, PowerBI, or another platform. But in that process of writing calcs and plotting measures, we check our work as we go. We know better than to push a dashboard into production without doing a QA pass, sometimes with multiple levels of review.

Maybe we need to dial down the polish. Prototypes that appear hand drawn can help communicate that the data represented is imprecise, uncertain, or approximated, which can be done with a number of different approaches. Perhaps we should ask for more sketchy renderings of dashboards in Claude instead of pixel perfect designs.

Then, we need to double down on prioritizing review and quality control.

When a designer can go from an idea to pushing a vibe-coded dashboard to production on a website in less than a day, I expect even designers assume that the AI tools did the math correctly or don’t have access to the source datasets to spot check against.

Slowing down also requires expectation setting with our users. We need to be thoughtful about getting our clients and data-adjacent colleagues on board with pausing and reviewing when they’re learning to expect polished results even faster—and perhaps trusting genAI tools with more math than they should.

Remember, all it takes is one bad number for a team to lose trust in their data. Before we ship something, it’s our job to ensure charts are accurate and check the text components, like numbers, calculations, and annotations. Which might mean you spend more time quality checking the dashboard than it took to spin up the dashboard in Claude, given how quickly it can create a new interface.

A moment for empathy around the black box

I see this moment as an opportunity. All of us who have been hand-chiseling charts since before the LLM explosion can be thoughtful coaches for colleagues and friends. We can help others continue to be careful creators, readers, and amplifiers of charts, especially when the stakes of getting the numbers wrong are high.

Years ago, graphic user interfaces made it possible to create charts from data tables without writing a line of code. I benefited from that shift, since I’m not a coder, and the tools sped up the process of chart design and dashboard creation for millions of others. Opening the door to data viz to more people is a good thing, but I’m sure we could find coders concerned about the mistakes that people would make when it became so easy to build a new chart.

genAI tools are removing another technical barrier to data visualization. When a person can prompt their way to a good looking chart without actually touching a data table, we can expect more people will become chart creators.

There’s a lot happening in the background to get from prompt to result that feels a bit black-box though, even for experienced designers. Ask question, get summarized answer. Ask for a bar chart of a particular data set, get bar chart. Ask for a chart to be enhanced with multiple complementary metrics on a dashboard, and you can have something interactive in no time. Most of us don’t have the deep knowledge of how these models are creating their outputs; we just review and assess what we see.

For some of our stakeholders, our work as data people has been the ‘black box’. We take their requirements and query data, run calculations, spin up charts, check our work, push to production, and ask our users to trust that what we present to them is correct.

Perhaps this moment is an opportunity for empathy, as designers experience some of that same black box sentiment about new AI tools that our colleagues have felt in the past about our own work.

And then we can help as the humans in the loop, saying like a broken record, “have you checked those numbers?”

Share

If you’re a public health or non-profit data person who wants to level up your charts (whether you’re hand-chiseling in Excel or prompting in Claude), there’s still time to sign up for my upcoming workshop on Data Visualization for Public Health, now scheduled for Friday, June 12. Use code ‘vizresponsibly’ for a big discount on enrollment!

Next month is packed with events that create space for the human-to-human conversations that feel more valuable now than ever. I have more thoughts on the ways the AI fervor (more! faster! don’t get left behind!) is reshaping how I spend my time, but will save those for another week since there are some registration deadlines coming up for cool data things.

Join us for Chart Chat 70

Happening tomorrow (April 30): Chart Chat 70! The four of us will talk through redesigns of a very decorated bar chart. It’s one of our fan-favorite episode structures, where you get to hear about our thought processes redesigning a chart and learn some of the tiny tweaks that make a difference. We don’t always agree on our design decisions, and that is a good thing.

And if you haven’t bought Dashboards That Deliver (the book I co-authored with my Chart Chat co-hosts), now’s a great time to ‘add to cart’ with one of the lowest prices I’ve seen on the book (~$18 on Amazon).

Coming up in May…

May 15 - Data Visualization for Public Health

A virtual public workshop on Maven, packing in my top tips for designing better charts in a 2.5 hour sprint workshop. Real examples using public health data, activities to practice, and a small group setting where you can ask your sticky questions.

Register today using the code vizresponsibly for 50% off the registration price. If cost is a barrier to joining and you’re a public health professional between jobs / in the public sector / in nonprofit work, reach out and we can chat.

May 19-20—Dreamin’ in Data

I’ll be in Chicago, IL to deliver a new talk on how Visualizing Data Responsibly is a Team Sport, especially in an AI obsessed world.

You can also find me Wednesday on the keynote panel featuring women leading in AI and data. I’m excited to join some phenomenal women shaping this field for a candid conversation, and will be around for the whole conference to connect with other data leaders in Chicago. Registration is still open for a limited time.

May 29—A Private Data Visualization Workshop

I’m heading to Maine to host a private data viz workshop for a group of library staff, nonprofit managers, and health care folks at the end of the month. Over the course of a day, we’ll tackle some of the fundamentals of data communication, chart design, and data storytelling with plenty of hands on activities using a real case example and data provide by the host organization.

In-person workshops are still one of my favorite ways to teach and learn, especially for a crew who have such an impact in their communities through their work. I always look for ways to customize examples and exercises with data relevant to the attendees, and leave plenty of space for Q&A.

If you’re making plans to level up your team’s data communication skills in the fall, I’d love to chat now, as my calendar is starting to fill.

Looking forward to June

Early bird registration for Outlier, the annual conference hosted by the Data Visualization Society, is open now! DVS Plus members always get the best rates on the conference and other events, but the early bird is a great option if you’re not ready to commit to membership.

The theme this year is The Final Draft with a focus on iteration, collaboration, and the messy in-betweens we’re often grappling with in data viz design.

The team has been hard at work behind the scenes curating an agenda packed with opportunities to connect, network, get resume and portfolio feedback, and more in addition to an incredible main stage line up.

The conference is fully virtual this year, which means we’re putting all of our energy into making the best virtual experience possible. Grab your ticket and make sure you don’t miss out!

I know that’s a long list of ‘where to find me’ that diverts from the usual data musings and charts that I feature in this newsletter. I’ll be back to the regularly-scheduled content next week! In the meantime, I’d love to hear from you if you’ll be joining the upcoming workshop or attending Dreamin’ in Data next month.

I delivered a talk at UW Milwaukee’s Zilber College of Public Health, saw the Art in Bloom installation at the Milwaukee Art Museum with my mom, watched my son’s class light up a stage in their Friday production of Pajama Party the Musical, and then watched floral dresses walk a runway with friends on Friday night.

I usually write about practical data visualization things in this newsletter, or very obvious ways we engage with data in the world.

But today, after coming out of a week of celebrating art and a weekend of spring yard clean up only to open LinkedIn to see posts about the new Claude Design release and how it’s going to help us prototype and ship dashboards faster…I want to celebrate slow, purposeful joy that happens away from a computer.

Subscribe now

Finding joy in data experiences

One of my favorite activities to start a data viz workshop involves flip charts, stickers, markers, and post it notes. I’ll ask people to ‘plot themselves’ as they get settled and grab a coffee. Flip charts have the shell of scatter plots and dot plots, often with very relatively scaled, imprecise axes, like data newbie to data expert. In the minutes before folks settle in to kick off the workshop, I take a peek at the charts.

The exercise is one of my favorite openers because people participate in data collection and chart creation, conversations start between workshop attendees as they plot themselves differently, and I gain insights about my workshop participants.

The data world has some exceptional creative talent who craft far more sophisticated data experiences than scatterplots on flip charts. From Dorsey Kauffman’s Ripple Effect exploring water pollution to Rahul Bhargava’s Data Sculpture 1659 (pictured below) sparking conversation about food insecurity, installations create ways for people to experience data, often away from screens and in real world settings.

In my talk last Thursday (The Human Elements of Health Data Visualization in an AI Obsessed World), I dove into a few different examples of data experiences and the creative ways we can help people connect with data.

The art pieces are brilliant on their own. But even better are the facilitated experiences and conversations that happen because these pieces exist in the world.

They often prompt questions and stoke curiosity around what they represent and why’ve they’ve been created. And when so much of the world seems to be rushing to create more things faster, the handcrafted aspect of these installations brings its own sort of joy.

Taking inspiration from Art in Bloom

Art in Bloom installations happen across the country, where floral designers take inspiration from art to craft complementary arrangements.

Some feel like literal reimagining of the colors and components in the art through flowers. Three bright yellow spots are repeated as big yellow orchids, a girl in a blue dress is represented with a cluster of delphiniums, or large white stones in a sculpture are represented with massive white proteas.

Others take more creative interpretations. ‘White Alice’, a washed out Alice in Wonderland piece, reimagined entirely with a rainbow array of bold flowers bringing to mind the idea of a wonderland. A monochromatic orange piece interpretated as various orange and coral blooms with layers of acrylic for depth.

Even more remarkable than the arrangements were the dresses on display Friday night, either crafted by a designer with a coordinating floral component (like a headpiece) or covered completely in live plant materials.

The surprise and delight from the floral creations came from artists reimagining art through new mediums. Flowers for spring? Groundbreaking…if they’re live floral gowns.

What do flower arrangements have to do with data?

First big picture, the zoom and filter, then details on demand.

That’s Ben Shneiderman’s information-seeking mantra, from a paper two decades old that still serves as a guiding principle for dashboard designers.

At Art in Bloom, as you view an arrangement from a distance, it’s easier to see the designer’s vision of how a flower arrangement ‘recreates’ a painting in its own way.

Step closer and you can identify all of the individual blooms and foliage being used, focusing on some of the unique plants used to bring a vision to life (lots of my favorite pincushion proteas were in this year’s displays).

Then if you really want to know the names of all of the flowers and the inspiration being the art, you can read the pedestal, which gives you all of that detail written out in text.

At the fashion show, the models walk down the runway and you first see the design from a distance and then closer up. After the show wraps, models and designers stick around so you can see the details up close.

First big picture, then zoom and filter, then details on demand.

Even on simple charts, we have to think at both levels: the first impression and then the details. Tinkering with a design, we make many decisions at the pixel level, removing clutter and making the marks sized appropriately.

But often our audience is first viewing our chart through a zoomed out lens: what patterns jump out from the data before they ever read a legend or the tiny text labels? What might draw them in for a closer look?

I don’t think our understanding of a chart has to be instantaneous, but the power in data visualization is that we help people see patterns in information before they even realize their mind is drawing certain conclusions. That applies to installations and data experiences, and to the charts we see in the news and on slide decks.

Take a pause to lean back away from your screen next time you’re creating a chart. What patterns jump out? How can you help the reader focus on what’s most important?

And take some time to step away from your charts all together and seek out something beautiful in the world, away from a screen. You may find surprising sources of inspiration, or just have fun. Either way, it’s time well spent.

If you’re a public health professional looking for ways to invest your screen time learning how to create better charts, I’d love to have you join my upcoming Data Visualization for Public Health live workshop happening May 15 on Maven, a platform doubling down on the value of humans-teaching-humans.

We’ll focus on practical, immediately actionable ways you can improve your charts with activities and discussion using public health examples (not Superstore datasets). Bring your questions and sticky data viz challenges! As part of the workshop, you’ll also get a recommended list of my top resources for continuing your learning journey.

Subscribers can use the code ‘vizresponsibly’ to get 50% off the registration cost. Register today and send me a note about what you’re excited to learn!

Subscribe now

What excited me about public health was the focus on policies, programs, and health systems that could help keep communities healthier, instead of focusing on treating the patients who come through my office each day.

But despite the fact that public health programs and policies touch our lives every day, public health success is often invisible. Success means better long term outcomes (think improvements in life expectancy) and fewer bad things happening.

What does that look like in practice? You can thank public health for supporting:

• the widespread adoption of vaccines to reduce morbidity and mortality from vaccine preventable disease,

• policy change requiring seat belts to reduce car crash fatalities,

• food safety regulations so you can eat bagged salad with (relative) confidence,

• the rapid responses from epidemiologists stopping outbreaks before they can spread

And so much more.

Public health runs on data.

Across specialties, many public health practitioners use data daily in their work. Those invisible successes I just listed? All supported by data.

Vaccine adoption is driven by communicating data about effectiveness and safety in understandable ways to patients and parents.

Policy changes are informed by data that helps us make the case for requiring wearing seatbelts when we observe that they reduce crash-related fatalities.

Data collected from routine food inspections or reports from consumers about illness are necessary to identify the offending food items and batch numbers for recalls.

Monitoring data for reportable diseases at national and international scales help contain outbreaks.

The list could go on. Public health professionals across specialties use data to inform policies, program decisions, and even communication campaigns. Even for practitioners outside of epidemiology and biostats, where there’s a clear focus on data as core to the work, effective public health teams include data capabilities.

So, in celebration of National Public Health Week, here are ten reasons (in no particular order) that my public health colleagues are some of the best data people I know. We may not always make the prettiest dashboards (as evident with some of what we saw during COVID), but actually using data thoughtfully trumps aesthetics.

Ten reasons why public health people make amazing data professionals

1. We’re systems thinkers, and can shift our focus from one hospital or community up to national and international policy setting. We see how one unit fits into the bigger system, and how data issues at one site can impact overall data quality as information moves through a reporting structure.

2. Evaluating how data is collected is part of our training. We don’t just take data sets at face value. We ask about methods, sample sizes, who funded the work, and other details that can get missed when you just ingest data tables into your analysis tool of choice.

3. We work with diverse data sources. We often have hands on experience conducting surveys, working with state or national statistics offices, and sourcing secondary datasets.

4. Context is key. We seek out additional data and information when assessing the scope of a problem or proposing a solution, and add context when communicating data. (Perhaps sometimes adding too many caveats in an analysis, but we can’t be perfect.)

5. Our love for accurate denominators knows no bounds. We care deeply about population sizes and getting those numbers right, since they help us understand the burden of disease and who we need to reach with key information. We’re always going to ask ‘what’s the n?’ when we see a big number.

6. Seeing the world through social determinants of health means we’re more likely to ask for the underlying distributions behind averages and medians and look for influences beyond the obvious for why a group has poorer health outcomes.

7. We’re scrappy. Public health isn’t one of the most well-resourced fields, so we can often make data magic happen with the tools we already have, and do it quickly.

8. Maps are meaningful, not just pretty pictures. We know in the US that zip codes are one big predictor of health outcomes (as a proxy for many other social determinants), understanding access to care requires geographic data, and viruses move through adjacent populations without any regard for national boundaries.

9. Social and behavior change communication is a core skill, which equips us with frameworks to think through how data is communicated. It’s not just about sharing a number, but about finding trusted messengers, putting information in the right place and at the right time, and giving the level of detail needed to inform without overwhelming.

10. Ethics are core to our work. As Bill Bicknell said when he opened our health policy course, “Public health is the art and science of deciding who dies faster and with less dignity.” Our policy, programming, and budgeting decisions have real impacts on human lives, and we’re acutely aware of that fact.

Happy National Public Health Week to all of my colleagues across this field. It’s been a rough year+ for us here in the US, but the challenges also create new opportunities for us to amplify and support each other.

If you’re in Milwaukee, WI, I’d love to see you at a lecture I’m giving next week at the UWM Zilber College of Public Health about The Human Elements of Health Data Visualization in an AI Obsessed World. The event is also free to stream!

Subscribe now

March 31: How is genAI changing how we teach?

Tomorrow, Jeff and I are going to pop up for a short Chart Chat about how genAI is changing how we teach university courses on data.

My experiences are more specific to teaching data visualization at the Maryland Institute College of Art in the Data Analytics and Visualization program, while Jeff teaches across data and tech topics the University of Cincinnati in their business school.

While the obvious question many people ask is, “What genAI tools are you teaching students to use?” the more interesting question for me has been, “How does the availability of genAI tools reshape instructional design?” i.e. how we teach, reinforce learning through assignments and projects, and assess competency.

You can join us live on YouTube or LinkedIn, and add your questions and reflections to the conversation.

April 16: The Human Elements of Health Data Visualization in an AI-obsessed World

Next month, I’ll be at the University of Wisconsin-Milwaukee Zilber School of Public Health as the spring guest for their seminar series on April 16. My talk focuses on what makes public health data visualization unique, particularly as technology creates more distance between us and the underlying datasets.

As genAI tools reshape how many data visualizations are created, health data requires more than just a human in the loop. Often, we see ourselves or people we care about represented in those data sets, sometimes from exceptionally challenging periods of illness or frustration navigating health care systems.

The human elements of health data require that we handle data with care, taking into consideration:

• Who is represented in the data?

• Who is the messenger sharing the data?

• How data may be (mis)understood and influence health behaviors?

The lecture is free to attend and open to the public to join in person or virtually—details here.

Bonus: Why Visualizing Data Responsibly is a Team Sport at Dreamin’ in Data (May 19-20, Chicago, IL)

The shape of data teams is changing. Budget cuts seem to be forcing more people to be ‘data unicorns’ at the same time that new AI tools are coming into the mix with the promise to simplify different parts of the design process.

I’m speaking about the teamwork (across people and tech) necessary to make sure we’re visualizing data responsibly at the Dreamin’ in Data conference in May and would love to see you there.

Registration for the two day conference is only $300 ($150 for nonprofit folks) through May 3. If your company provides professional development funding, the organizers have some great resources on the event website to make the case to have your costs of attending covered.

If you enjoy reading Viz Responsibly or someone forwarded this newsletter to you, I’d love for you to subscribe so you don’t miss an issue.

Subscribe now

“You can generate SO MANY ideas, SO QUICKLY!”

“I made this app in an afternoon!”

There’s a lot of commentary these days about how quickly you can make data applications. It’s prompted questions about the role of data viz developers in a future where anyone can prompt and deploy a new dashboard without opening Tableau or writing a line of code. (Note: I think those capabilities when working with complex, sensitive, sometimes poorly organized enterprise data, are overstated.)

One of the selling points of the newish cohort of genAI tools is enabling users to create things quickly. That includes lists, text, charts, and even fully deployed applications, like we’re seeing as outputs of experiments with Claude Code.

Many of the experiments I’ve seen are for some an audience of one, or broad tools designed for anyone to engage with. Some of them are very cool! But when we’re designing charts, infographics, and dashboards for specific goals, we can’t rely on our own preferences.

If you’re building a new dashboard where the only back and forth is between you and Claude, even if you work through various prototypes, you’re likely to miss out on insights that would reshape some of your design decisions because you aren’t getting input from other people who will be future users of your tool.

And I worry, that in the drive towards ‘speed to deployment’ as a marketing promise and a business priority, we’re going to make insights from data less usable as we:

• Push users out of the design loop,

• Miss out on insights on how people actually engage with information, and

• Set ourselves up for having more data, but less engagement thanks to information overload

Our biggest issue today isn’t not having enough information. It’s making sure people have meaningful information: the right data, at the right place, at the right time, and in a meaningful format.

Let’s unpack three pitfalls happening when data design collaboration is between one person and a machine, and recommendations for data communicators to avoid these pitfalls in their work.

Pushing users out of the design loop

Remember human centered design? User centered design? Design thinking? All of those methods center on engaging our audience members as partners in the design process. And we shouldn’t shelve them entirely in favor of machine-centered design just yet.

In Dashboards That Deliver, our design framework is a loop that starts with a spark and a discovery phase to define the needs, and returns back to revisit if the dashboard is meeting those needs after deployment.

The Discovery and Prototyping phases are shaped as a ‘double diamond’ in homage to the concept from the British Design Council, where you want to gather many insights and then focus on the specific problem you need to solve; then, you brainstorm many possible solutions and focus on one concept to take into development.

A great discovery phase doesn’t just ask people what charts and filters they want. A great discovery phase includes conversations that uncover unexpected insights about how and where people engage with information, what their current pain points are, and more. The challenge is that takes time, from scheduling the meetings to synthesizing the results, and the world seems to be growing increasingly impatient.

But when we pause and take that time, we can learn about the data needs of people not just a list of functional requirements, generalized based on job titles.

What that means for data people in our AI obsessed world:

If you’re building a data application (anything to explore or explain some kind of information), plan for a discovery phase.

• Name it. Put it on the roadmap.

• Identify who the critical users are that you need to learn from, and then take the time to have open ended conversations.

• And at the end of that phase, write up the key insights, and be clear from the start that there will be outputs from this phase that your team/client can see.

Even if you’re using genAI tools as part of your design process, taking the time to listen to your audience first, synthesize what you hear, and then using those insights to form better prompts for brainstorming or directing your tool of choice will create a better output.

Missing out on how people actually engage with information

What might get lost in the seemingly simple dashboard framework is just how often designers who deliver user-friendly dashboards talk to users. It’s not just during the discovery phase.

When I’m creating data visualizations, those moments can include:

• Interviews and group conversations during discovery

• Requests for feedback on mockups and wireframes in prototyping

• User feedback sessions on working versions of the dashboard during development

• Engaging users as champions for the dashboard, encouraging use by their team members, after the official go-live date

• Capturing feedback in structured ways to plan for future enhancements and changes

The same large group included in the discovery phase aren’t all consulted at every step. But each touch point creates an opportunity to build a tool aligned to your audience’s needs and (perhaps more importantly) for your users to feel some ownership over the final product. They’ll have insights into why certain features were included (or not) and get excited about the components they influenced.

You can certainly bring genAI tools into the process, but show what you’re creating with your machine partner to your users for their feedback too. Engage your audience as partners in the process, and you’ll have a better final product and build champions along the way.

What that means for data people in our AI obsessed world:

• Identify up front how you want to engage your audience throughout the design process, and share those expectations with them.

• Define their role as a partner, not just a customer, while also setting boundaries around how prioritization decisions will be made.

• Show them messy work in progress moments and A/B decisions to be made, ask for their feedback, and dig deeper than ‘which do you like better’ and probe ‘why’.

More content, but less engagement

A World Bank study from 2014 looked at downloads and citations of policy reports—written for the purpose of shaping policy with data and insights—as proxies for engagement and use. Their central findings:

“About 13 percent of policy reports were downloaded at least 250 times while more than 31 percent of policy reports are never downloaded. Almost 87 percent of policy reports were never cited.”

The study is more than a decade old, but the challenge of knowledge being pushed out into the world but not reaching its audience remains.

There’s no shortage of charts or data tools at our fingertips. We see them at work in slide decks and dashboards, in newspapers, and on social media feeds. But there is a certain information overload these days that can cause many people to just tune out, and a trust gap where we seek information from the people and sources that we know we can rely on.

AI tools can build a data product with a series of prompts. But is Claude going to pop into someone’s office and ask if they’ve tried out the new dashboard before the next monthly meeting where the team reviews their progress?

Having a real person nudging their team to look at a new set of charts or try out a new dashboard is one of the best ways to support engagement with what we build. When we engage our audience as partners in the design process, we create those kinds of champions and amplifiers for our work.

What that means for data people in our AI obsessed world:

• Set expectations from the start that you’re looking for people to help spread the word about your new dashboard or data tool when it’s live. Don’t just assume that publishing something to a server or the web will mean it gets used.

• Make a communications plan that goes beyond one email. How are you going to spread the word that your new dashboard is available?

• Measure success and celebrate together. Whether you’re tracking page views or ways that a dashboard shaped decisions, be clear on what you’re measuring and share your wins with the people helping you get the word out (along with a note of thanks for their support).

Let’s build better, not just ‘more’ and ‘faster’

AI tools are evolving faster and doing more than most of us expected. But just because we can create things faster, doesn’t mean we should.

The conversations that happen through the design process across the team and with the audience do more than just help us decide if something should be a bar chart or a line chart.

Through the process, we build trust in the data, improve overall usability, and build enthusiasm for sharing the final product, all of which make the difference when we’re looking for data to be used, not just published.

Note: I want to acknowledge that there’s a whole new horizon where the end users are machines, particularly for ecommerce where agents can be deployed to make purchases on behalf of a customer.

If someone’s agent lands on your site to make a buying decision, how pretty your website doesn’t matter. What counts is how easy it is to gather information from the underlying data layers. As data people, it’s worth thinking through what it means to design with a machine as the customer, as there may be a point when people are seeking information and data insights using agents. That’s still a flavor of design thinking though, just focusing on agents as customers.

has a brilliant book and MX Strategy Map that are useful primers on this new horizon.

I don’t think I’m alone in that sentiment.

And I’ve wondered what that means for getting eyes on data that can shape perceptions and decisions, from news events to getting people to engage with dashboards.

Jon Schwabish wrote a newsletter two weeks ago about the changing media consumption habits that could make public dashboards obsolete.

I disagree with that absolute sentiment: dashboards certainly aren’t dead, even if tech platforms have taken to giving them different names like liveboards or algorithmic cockpits ← not kidding, this one surfaced in our research for Dashboards That Deliver.

But I have observed across my career that you cannot assume if you build a dashboard, people will use it to find the information they need.

Today, with limited attention spans, people are looking even more for curation, key takeaways, and guided analysis through a chart or series of charts. Data storytelling has gotten its flowers for a reason—it emphasizes the narrative, text components that demystify how to read a chart.

More of those data stories works well when we have key takeaways to share. But sometimes there really is a need for data exploration, and thinking bigger than our laptop screens.

I’ve been thinking a lot about physical data installations as part of my prep for an upcoming webinar with SkyKit about extending dashboards to drive frontline action by displaying them on big screens in health care and other industries. I’d love for you to join the conversation! It’s free to attend, but you need to register in advance for the event details.

What do our limited data attention spans mean for dashboards?

The first principle of good dashboard development is more important than ever: you need to have an audience and purpose for your dashboard for it to be worth creating.

No, ‘the general public’ isn’t a well defined audience. A dashboard designed for everyone is a dashboard that meets the needs of (almost) no one. For me, those are the public dashboards that often ‘fail’ despite requiring time and investment.

Yes, you may be really excited about your survey findings, a new data set, or even a whole suite of project metrics. But you need an audience who brings that same motivation to the table around engaging with the data.

Even with a defined user group, we can’t assume enthusiastic dashboard users will explore all of our interactivity and filters. For designers, that means we need to:

• Be a ruthless prioritizers in our design decisions: the most important, actionable data needs to be front and center, not buried in the fourth page of the display.

• Personalize automatically: Where we can, personalize the dashboard automatically (based on location, user group, etc.) rather than requiring action.

• Bring the data to where people are: Give more thought to where the dashboard is being viewed and any friction points to engaging with the data.

The first two items are generally good design advice. But dashboards out in the wild and on big screens? Doesn’t that sound like a data security nightmare compared to individually provisioned logins?

The value of physical data displays

Yes, we need to consider privacy and security in how data is shared.

But putting aggregate counts and numbers into the spaces where people are working (rather than only making them available when they’re on a computer, logged in, and clicking on dedicated links) has been done long before we had mega-sized screen displays in our offices.

Let’s peek at three examples of physical data displays with an impact.

Wall charts for immunization data

I worked for years in global health, where our approaches to data visualization were often scrappy and low tech, but highly effective.

Take the Immunization Monitoring Chart, championed by the World Health Organization and Ministries of Health around the world. The chart allows all staff at the clinic to view aggregate statistics on immunization coverage, and often includes context around how to interpret the chart and what to do with the data. Simple, but effective to make the data more accessible and transparent.

Making data collection part of the installation

Jason Forrest and Jen Ray are the creative geniuses behind the Data Vandals, which creates installations that integrate art and data in novel ways.

(Jason is also the visionary behind Nightingale, the Journal of the Data Visualization Society, and leading the new Data Communication & Visualization masters program at the School for Visual Arts in NYC. They’re accepting applications for their first cohort now!)

For the Great London Survey conducted by the Data Vandals, people added their perspectives as data points (stickers) on scatterplots to share sentiment about the city. While you’re not generating a computerize data set with this sort of data collection, you’re making participation more accessible, engaging, and fun.

Creating a mission control center for Planet Earth

On the other end of technological complexity, there’s one of my favorite dashboards in Dashboards That Deliver: the NASA Earth Information Center.

The team described the original concept for the installation as a ‘mission control center for Earth’, showing real time satellite data about various climate measures. The final installation has a 20 foot long hyperwall that loops dashboards, video segments that humanize the data stories, and ‘beauty pieces’ that function as data art brain breaks.

The original concept wasn’t just about showing data: it was about creating space for conversations. While the data is interesting, in a world where climate science has become a politically polarizing concept, looking at a dashboard together and then having a conversation around what we’re seeing is where the magic happens.

Bringing data to the people

Technology keeps changing. But there’s an evergreen need for information across industries that requires curiosity about how to position information in novel spaces where people may be more likely to pause to pay attention.

We need to think about how, where, when, and why people engage with our data communication tools, including dashboards. In big public health projects, my work was never funded by a dedicated ‘data visualization’ workstream—and that was a good thing. Dashboards were just tools to make data more accessible and usable by people.

If we focus only on the tool, and not on the people using it, we will miss opportunities to position data such that people have the right data, in the right place, at the right time.

And in an AI-obsessed world where people are collaborating with digital assistants and chatbots more and more, we need to create real world spaces for human-to-human conversations.

Maps have very practical use cases, particularly when proximity matters (disease outbreaks) or when they’re part of larger geospatial analysis projects analyzing borders, distance to care, and more.

But that love for maps isn’t always rooted in needing to see data geographically.

So let’s unpack why people seem to love maps, how I approach client conversations about maps that don’t seem to hit the analytical mark, and three of my favorite charts to pair with maps to help show patterns and rankings more easily.

Subscribe now

Why people love maps in public health (and beyond)

I remember working on PEPFAR data communication products (dashboards, slides, etc. ) where a map of HIV prevalence by country or some other single-variable choropleth map was required by the team. The map made a great intro slide to show the reach of our programs or was a graphic they were used to having on hand.

But what ah-ha moments did those single variable maps usually create?

On a global scale

Let’s take a look at a world map of maternal mortality, in the spirit of unpacking a real world example. This one is from the World Bank.

What does this map tell us?

• Relatively speaking, maternal mortality is higher in sub-Saharan Africa than most other parts of the world. Bonus that we’re looking at a pretty familiar map (the world!), though the specific projection being used emphasizes countries that have large land masses. (Tile grid maps can help address this distortion.)

• Afghanistan is an exception, which jumps out with the darker blue shading.

• Most countries are under 62 maternal deaths per 100K live births, which is apparent when we peek at the color legend.

What doesn’t this map tell us?

• Specific maternal mortality ratios, unless we hover over a country to see the value on the tooltip, because country rates have been binned into five colors on the legend.

• Insights around the precise rankings of maternal mortality across different locations.

• What to do with this data—we’re not seeing a connection back to the other metrics that could help us understand why maternal mortality is higher in some countries—or disparities within the countries.

Going local (to me)

Let’s look at a state level map, where the geography (counties) might feel less familiar than a world map. I live in Wisconsin, where we’ve had alerts for multiple cases of measles pop up this year, as this vaccine-preventable disease makes a resurgence.

Being somewhat new to the state, I was curious about vaccination rates across counties. Counties close to where I live are of particular interest, since they could impact the spread of measles near me. Seeing the data on a map helps in this case, since I don’t know the names of all of the counties surrounding Milwaukee, but I can see them on a map.

The Wisconsin Department of Health Services has a number of interactive data explorers about immunization data, which include a map for coverage of the measles-mumps-rubella (MMR) vaccine.

Like the World Bank, they’ve binned the color legend (here, 10 sequential colors with darker showing higher coverage). I love that they’ve paired this map with some added context where I can select a county of interest and see how that county stacks up against the state and view current stats about Milwaukee.

But like the World Bank map, this county map is at best allowing for relative comparisons (highest—darkest and lowest—lightest), not precise rankings of the counties.

Plus, it’s missing perhaps the most important piece of context: how are these counties doing compared to the target coverage needed to prevent the spread of measles?

You may remember from COVID that we call that herd immunity. For measles, we are looking for 90-95% coverage to stop the spread of the virus since it’s exceptionally good at moving through unvaccinated populations (Hijano et al, 2025).

How I approach client conversations about maps

Both of the maps we’ve unpacked so far could be interesting to audiences who are curious about maternal mortality globally or vaccination coverage in Wisconsin. But at best they facilitate relative comparisons between geographic units.

So what do I do when a client says they need a map, but they want to make more precise comparisons?

1️⃣ I clarify why and for whom. Who is the audience for this visualization? What do they need to do with the data? Are they looking for visual variety in a deck already filled with bar charts? Are they used to having a map for this data (even if it’s not often used)? Is geographic proximity of data points important?

2️⃣ I offer alternate options. If the map seems to miss the analytical mark connected to the ‘who’ and the ‘why’, I mock up alternates. Ranked bar/column charts, beeswarm plots, and dot strip plots are options I use most frequently (more on those below).

3️⃣ Gather feedback and listen. We talk about what works well for the different visualization options and make a decision that reflects their needs not just my professional preferences.

Sometimes, the answer is that the map stands alone, particularly if it’s in an interactive format that lets a user see the details they need. But often, we end up with some kind of dynamic duo of map + chart that addresses the client’s needs that we’ll unpack next.

Three of my favorite charts to pair with maps

Beeswarm plots

Beeswarm plots display an organized ‘swarm’ of points, with one mark for each geographic unit. Points don’t overlap, instead having the cluster of dots get wider or narrower along a central axis.

To add to the readability, we could label the highest and lowest value or add color to specify if counties are above or below a target—here, all counties are below the 90-95% herd immunity target though. Instead, we could consider a state average line for comparison to see how a county is doing compared to the state.

(And because I often get asked: I create beeswarms in rawgraphs.io. Then, I edit the downloadable svg in a design tool like Figma or directly in Powerpoint by converting the svg to shapes.)

Dot plots

Like beeswarms, dot plots display one mark for each geographic unit.

The dots can overlap along a central axis, sometimes obscuring each other if the values are the same. We can ‘jitter’ the dot with some randomness along an invisible y-axis, and while the marks may overlap we can see where they cluster a bit more clearly. Making the marks semi-transparent helps make sure marks aren’t blocked from view entirely.

Dot plots are particularly useful on dashboards or interactives, where we can hover over a county on the map and see where it ranks along the dot plot (including how far it is from the target, here plotted as a reference band). Bonus points for adding in a human-readable tooltip that describes what’s happening to add some extra context for the user.

Ranked bar or column

If you have more space on the dashboard or slide, a ranked bar chart is one of the most familiar to most readers and can allow every point to be labeled for quick reference.

The labels are practical on static displays, like on slides, but the space this chart requires can be a challenge when you want to keep the county names displayed. Solves that squeeze in all of the labels can feel cluttered, and require rotated little text or limiting the chart by region.

Similar to our dot plot, we benefit from interactivity here to highlight where a county falls on the bar chart.

The big takeaway: make sure your map is delivering the analytical insights your audience needs, and consider pairing maps with other charts for quicker speed to insight. There isn’t one right answer, but it’s worth a conversation to see if a map alone can do the job.

What are your go-to charts to pair with maps? I love seeing examples of these kinds of dynamic duos of charts out in the wild!

I peeked at the date on my last post and cringed a bit: September 2025. Not the writing pace I had set for myself when kicking off this newsletter, but there’s no time like the present to do a reset.

From updates over on Chart Chat or LinkedIn, you may have seen…

• Dashboards That Deliver was published in September and we’ve loved seeing the ways the book resonates with data viz creators. It’s the book I wish I had when I started my data viz journey. The design framework is useful whether you’re hand-chiseling dashboards in PowerBI or iterating with a genAI tool, and I hope you’ll grab a copy if you don’t have one already.

• As part of the book launch, I did more than a dozen talks in Columbus, Milwaukee, Washington DC, and many virtual spaces, and ran a two day in person workshop on dashboard design. September, October, and November were busy. I’ve been adapting the workshop into a virtual course that will run in the spring on Maven (watch this space for registration updates!) and I’ll share a round up of a few of the recorded talks in a future post.

• I’ve been spending more time connecting with my colleagues in the public health world as policy changes from the current administration create setbacks and communication challenges. I’m still supporting the Data Visualization Society as an advisor and love getting nerdy about chart design with other data people, but I’m also looking to reconnect more with my public health world.

If you want to spent an hour of your morning getting nerdy about charts today, join us for at 11AM EST on YouTube. And watch your inbox for a deep dive on why I like pairing maps with a chart in tomorrow’s newsletter, which you can expect to see in your inbox or here on Substack every other Thursday going forward.

1. Gearing up for Dashboards That Deliver to arrive in mailboxes this month, with the official publication date next Tuesday, September 23.

2. Managing information overload in my professional world of public health.

One week to publication day

First, on the bright side: the book I’ve poured countless weeks into, along with my coauthors Andy, Jeff, and Steve, is finally going to be out in the world!

Dashboards That Deliver is the book I wish I had when I started my data viz career journey.

The first section walks through a human centered approach to data viz design that doesn’t just say “know your audience first.” We have tools, techniques, and lots of examples and anecdotes that dive deep.

The second section has 15 real world examples of dashboards that have made an impact, digging into the process and constraints around each of the dashboards.

Where else can you find examples from a professional racing team, NASA, and nonprofits all in one book?

We’ve outlined ideas at the start of each chapter explaining how and why the scenario could apply in other industries and scenarios, so hopefully any data person will find useful ideas in each chapter (or just be inspired!).

The last section tackles industry topics like genAI and data visualization, collaborating on dashboard projects, and more. You can find the full table of contents on the book website (just scroll down on the home page).

My ask to you: if you know you’re going to buy the book and can choose to order by September 23, that’s one of the best ways to show your support. (THANK YOU to those who have already pre-ordered!)

Pre-sales (everything before publication day) and first week sales are what put books on fancy lists and nudge the algorithms to recommend the book to others. You can purchase with your local bookstore on Bookshop, or order through a company like B&N or Amazon (if the prime shipping still has you hooked).

And if you read the book, I’d love to know what you think. Send me a note, or share a photo on LinkedIn and tag me. Whatever suits you best! Or, come join one of the 14 (!) talks and workshops I have scheduled through October and say hi in person or at least in a Zoom room.

You can find the updated list, as registration links go live, on our website.

The information overload (most of it bad news)

The book launch has brought a lot of joy in the midst of the chaos of recent months. Professionally, the first half of this year was dominated by the impacts of the rapid shutdown of USAID. The second half feels more like a policy and staffing crisis in US public health with life and death consequences.

The appointment of RFK Jr. as the Secretary of Health and Human Services didn’t inspire much hope for what the rest of the year would bring for public health, but the past months have been particularly painful to witness.

From politicized appointments to the ACIP that oversees immunization recommendations to departures at the CDC with damning letters of resignation, including from Demetre Daskalakis, MD, MPH, director of the National Center on Immunization and Respiratory Diseases with this stand out line:

“I am unable to serve in an environment that treats CDC as a tool to generate policies and materials that do not reflect scientific reality and are designed to hurt rather than to improve the public's health.”

I’m grateful to some of the incredible science communicators who must be spending hours each day compiling content, including pre-bunking some of the misrepresentations of data flying around, particularly around vaccine safety.

A few of my favorites, all of which include simple visualizations with their content:

• - for weekly round ups and plain language explanations of what’s happening in public health

• Nini Munoz on Instagram, who has some of the best ‘will it correlate?’ charts I’ve seen (lululemon sales going up along with autism rates? what might we deduce…)

• for deep dives into specific topics and accessible carousels digging into how and why data gets represented different ways

• who breaks down scientific topics and shares personal stories about her experience growing up in an AV household. Her latest post makes the performative nature of the Make America Healthy Again agenda personal with her stories of food insecurity. (Also on Instagram if you prefer to read in graphics.)

And if you’re a real public health data nerd who also appreciates some pop culture references in a cheeky slide set, Eli Holder’s snip from his SIGGRAPH deck is 26 slides that paint a masterclass in setting up an argument about what’s broken in how we communicate (and can oversimplify) health data.

These days, in the midst of the information overload, I find myself going to the communicators and news sources that I know I can trust, and trying to do more offline reading.

Some of that in books and newspapers, but even more often seeking joy. Periodicals about SCUBA diving, fantasy genre audiobooks, and podcasts that make me laugh. Being informed is important, but for all of my public health colleagues (and advocates): we’re in a marathon, not a sprint, and need to hang on to a bit of joy too.

I hope Dashboards That Deliver brings some (nerdy) joy, particularly in the scenarios that celebrate federal work, like how NASA built their Earth Information Center, and public health data projects, like Johns Hopkins’ Coronavirus Resource Center and the familiar dashboard from the pandemic. Those are stories that deserve to be told.

I’ll be back to more routine updates now that the school year has started and book launch is finally here. Thanks for reading.

Using data in real life

Part of that curiosity stems from thinking though where we encounter data in our daily lives. I write a lot about charts in public health, data journalism, and other professional contexts, but the reality is that those are often for somewhat specific audiences.

More often, people will engage with data on wearable devices like an Apple watch, on weather apps, or when making decisions about getting from point A to B.

Consider travel, which requires a lot of decision making with numbers, even if you don’t think of them as data-driven decisions. Flight at 9AM? Just figuring out what time to get the airport, you might consider:

• how long it takes to get to the airport

• how long parking and getting to the terminal will take (if you’re not using a rideshare, train, or bus that drops you at the terminal)

• how early you prefer to be there

• whether you have bags to check or not

• how long security usually takes

• whether you’re taking a domestic or international flight

Then you calculate backwards from your departure time to sort out when to leave your house. Once you’re there, you relay on information on flight status boards, airline apps, and signage to make it to your gate in time.

Subscribe now

Three ‘just the numbers’ examples

While I love a great chart, the signs used for communicating information for people at airports are better suited to simplicity and often feature numbers or plain language to describe a number (see example #3).

Think about who the audience is: someone rushing by as they hike to their gate at the airport, not a captive audience reading a long article on the plane.

Here are three simple examples from my flights between Milwaukee and San Francisco and what they made me think about as a data viz designer.

Wait times

I love this board at the security checkpoint. What information do we need at that point in our journey? Some estimate that tells us if we’re on track to be through and to our gate on time.

Lately, I’ve seen TSA Precheck lines longer than standard security lines as more people get approved for the expedited process. But the length of the line doesn’t tell you time; Precheck tends to move faster since passengers aren’t unloading electronics and are often seasoned travelers. And we don’t need anything more complex here than a simple number.

Cost of repair

The moving walkway from the daily parking structure to the terminal has been out of commission for some time, and appears it won’t be back in motion anytime soon. Inconveniently, it’s the moving walkway towards the terminal when you may be in more of a rush, leading to some frustration for delayed travelers.

I love the transparency in this sign sharing more than just the status of the repair: to make the repair would cost $12 million dollars. That’s more than most people (myself included) would estimate a moving walkway to cost.

More visibility into our public works project costs can help cultivate some empathy for the cost-benefit decisions made every day that impact how we move through the world.

Transit times

Rushing between gates to make a connection or just trying to get from security to your plane?

When you’re walking along a terminal, the gate numbers often give a sense of how far you’ve gone. If I started walking at gate B10 and need to make it to B40, I can see the gates and how far I’ll need to travel.

But in hallways between terminals in unfamiliar airports? That can feel endless, which is a point of concern if you’re tight on time and need to know how quickly you should be hustling to make it to your gate in time.

I loved both the encouragement (“Nice!”) and the plain language time details (“You’re halfway there.”) on this wall decal between terminals. I’ve seen similar details on flight boards that show destinations, flight numbers, and the estimated walk time to a gate, but never quite know how my own walking pace will stack up to the estimates.

Signs help make anyone a data person

While complex graphics may be better suited for ‘data people’, signs like these help anyone understand and act on information. When someone tells me they’re ‘not a data person’, I love to chat about travel and weather apps, which help demystify data.

As a data viz designer, observing how we engage with information as we navigate the world is a valuable source of inspiration. Simple signs are also a reminder that we need to consider our audience and their information needs when we’re designing more complex visualizations. Where are people encountering our charts? How much time do they spend with them to glean insights? What information are we looking for?

Particularly when I’m working on dashboards, I find designers try to pack in all of the information possible across many tabs and views. But if we instead focus on serving up the information someone needs most, we’re more likely to help them along their way to a decision and create a more useful product (even if it shows less information).

Bonus: visualizing wine

The vacation destination last week was wine country. Imagine my delight when I found one example of data viz in the wild at our last tasting.

The signs above sat in front of each bottle at Frogs Leap. The signs communicate information about the flavor profile of the wine, identifying where the wine sits across the scales of:

• Taste - from fruity to savory

• Body - from light to rich

• Tannins - from smooth to tannic

What I love about these little continuum plots is the abstraction. We even use them as an example of a discovery tactic for making sense of how audiences are different in Dashboards that Deliver.

Instead of rating along numeric scales or with other units, they communicate information about the flavor profile of the wines between different common characteristics in wine. This could make it easier to compare across varietals and make selections, even if you didn’t taste each bottle. In contrast to the specific numbers at the airports, these were all about relative scales across bottles.

Even when I’m on vacation, I can’t seem to turn my data brain off. I hope you’re feeling inspired to observe more ways you see numbers communicated out in the world, and bring some of that inspiration into your own data communications.

Share

I’m enjoying more offline time over the summer with recent travel and quality time with my kids. As I mentioned in a recent LinkedIn post, hobbies that pull me away from screens and into the world of making things (like cooking) are bringing a lot of joy now, so this newsletter will continue be a bit sporadic through the summer months.

Share Viz Responsibly

Both were interesting, engaging talks and conversations. Both are topics that will shape the world of data visualization in the coming years.

But I felt oddly disconnected from those topics as my news feed filled with major shifts in geopolitics and, for me as a public health practitioner and parent with young children, new disruptions and means of sowing distrust around the effectiveness of vaccines.

This week, a newly seated members of the Advisory Committee on Immunization Practices (ACIP), met for the first time after Secretary Kennedy removed all of the previous members—an action he previously stated he would not take.

Gutting that committee raised immediate red flags for public health practitioners and clinicians, with the president of the American Medical Association stating:

“For generations, the Advisory Committee on Immunization Practices (ACIP) has been a trusted national source of science-and data-driven advice and guidance on the use of vaccines to prevent and control disease. Physicians, parents, community leaders and public health officials rely on them for clinical guidance, public health information, and knowledge.

Today’s action to remove the 17 sitting members of ACIP undermines that trust and upends a transparent process that has saved countless lives. With an ongoing measles outbreak and routine child vaccination rates declining, this move will further fuel the spread of vaccine-preventable illnesses.”

Science. Data-driven. These should be shared values when it comes to making health decisions, but seeing the sweeping changes and watching the ACIP meeting this week, we’re at a critical moment where people who cherry pick data points or misrepresent information are moving into more positions of power.

The messenger matters

Who sits on what committee might feel distant from our day to day lives, but the communications coming out of that committee and their decisions on what makes the agenda to discuss impacts us in ways you may not realize.

Health communications around highly polarizing and politicized issues can be complex. The ACIP plays a critical role setting recommendations for vaccines being recommended for different demographic groups. Removing scientists and replacing them with less qualified members pushes us further down the road of polarized, politically motivated decisions on a topic critical for our children’s health.

Data should give us a tool to communicate evidence based decisions, but we know it can be misrepresented. Sometimes that is done intentionally; other times, people don’t understand how to interpret study findings or makes sense of self-reported data in a database.

Emily Oster addresses six common concerns around vaccine safety, testing, and effectiveness on Parent Data, and even dug into VAERS data around the time that COVID vaccines were being approved to highlight how many ‘errors’ reported in the system were not the red flags anti-vaxx influencers make them out to be.

But who is reading Oster’s substack? I’d guess mainly parents already curious about the science behind different recommendations in pregnancy and childhood, which is what she built her public brand around with the books Expecting Better and Cribsheet.

Other science communicators frame data through the lens of their professional expertise and personal experiences. was raised in an anti-vaccine household. She shares data and evidence across different topics, but also adds her personal story including how her own beliefs changed, which can make changing beliefs feel more possible.

Science communicators play a critical role making data more accessible. But the broader public health and medicine communities are often sharing information in thoughtful, responsible, detailed, and overly technical ways.

Those can’t compete with anecdotes that strike fear in parents or misrepresent science with big numbers from VAERS that can scare parents, bringing us to the declining vaccination coverage rates and new outbreaks of diseases we see today.

Visualizing the impact of vaccines

Data visualization can play a key role in telling the story of how vaccines altered the trajectory of child mortality trends and the frequency of illness. And one of the reasons elevating effective graphics and continuing to share data matters is because the decision to vaccinate or not is one made by millions of parents each year.

I want to believe that parents are genuinely looking to make evidence based decisions in the best interest of their children. Many go to parent groups or blogs looking for information to inform their decisions, though I often remind parent friends that decisions about a child’s health care is best made in consultation with a pediatrician.

There’s plenty of misinformation about vaccines in the world, including made up or misrepresented stories to go along with the data. So I always keep an eye out for attention grabbing graphics where you can’t ignore the ways health outcomes changed after vaccines we introduced.

Vaccines have contributed in big ways to the sharp decline in child mortality over the last 70 years in the US and around the world, so simple line charts are often useful for communicating that change.

Around 1950, around 1 in every 25 children born died before reaching age five. Today, that number is around 1 in every 166 children, according to Our World in Data. Having experienced the loss of a child myself, I wouldn’t wish that grief on any family, which is a personal motivator for some of my own work in public health data.

Line charts are simple and easy to read, but they’re not always as eye catching as more creative pieces. My favorite visualization of vaccine effectiveness is from the Wall Street Journal, of all places, from a decade ago.

Each row represents one state, each column represents one year, and the color represents the number of cases of the disease reported in that period. A reference line shows the year the vaccine was introduced.

Here’s the graphic for measles:

You can’t miss that something dramatic changed after the vaccine was introduced. Cases declined and have been in single or double digits for the past decade+. And the novelty and bright color palette grab our attention in ways a line chart doesn’t.

Perhaps it’s a good time to revisit and update the data in the graphic, which would show rising case counts of measles in recent years. While we’re nowhere near 1960’s numbers, the US had 59 measles cases in 2023 and multiple nearly five times to 289 cases in 2024, according to Our World in Data.

The charts continue across multiple vaccine preventable diseases, all with similar patterns: cases decline significantly after the introduction of the vaccine.

The rainbow color palette sparked debate among data viz practitioners, but I stand by my choice: I love this graphic, and often bring it up when talking about stats on changes in disease incidence thanks to vaccines.

Remember, there are many issues around long term effects from these illnesses besides death, which is why focusing on cases and not mortality matters. Just ask a polio survivor.

Why this issue is urgent

I’ve had my own challenges as a public health professional and as a loss parent around engaging with antivaxx rhetoric and influencers, including having my own personal trauma repurposed and misrepresented to sow doubt about the safety of vaccines.

But this week, elevating evidence based recommendations around vaccines in childhood feels more important than ever for three key reasons:

• Messengers who question or outright oppose vaccines as a preventive health treatment, once considered to have fringe opinions, are now running agencies and staffing committees.

• Vaccination coverage is declining and these declines in herd immunity could bring us back to endemic measles in around 20 years, assuming vaccination levels remain consistent with current numbers, according to a study modeling the re-emergence of measles in the Journal of the American Medical Association. If you’re not aware, we don’t have curative treatment for measles, infection can have long lasting health effects, and it’s an incredibly contagious disease where you can be infected hours after someone carrying the virus has left a space. What does that mean in practice? My pediatrician was concerned enough about taking our son to a region with an active measles outbreak for a childhood friend’s wedding that she advised we skip the event since he wasn’t old enough to have his first dose of MMR.

• Removing recommendations impacts coverage and affordability. Yes, parents can make their own choices. But in America, where we rely on a mix of private health insurance, Medicare, and Medicaid to pay for health services, recommendations from groups like the ACIP are critical to having preventive care and treatments covered by insurance. Take COVID-19 annual shots, now not recommend for healthy children or pregnant women (who are at a higher risk of complications for themselves and their babies if infected). Shots could cost upwards of $150 per person if someone elects to get vaccinated but isn’t in a recommended group where the shots are covered by insurance. That’s going to put vaccination out of reach for many American families.

The terrifying combination of anti-vaccine influencers getting louder microphones and shaping policy, herd immunity declines that set the scene for more widespread disease, and adding structural barriers like increased costs to effective prevention has many of us in public health on edge.

What you can do

People listen to people they know and trust. Believe it or not, you can be a very effective messenger for sharing information, even if you’re not a data or health expert yourself. You can:

• Read, amplify, and share content from science communicators who use their free time outside of work as doctors, nurses, and scientists to create and share resources addressing common misinformation about vaccines and other health care topics. Some of my favorites include: @ScienceWhizLiz (linked above), @niniandthebrain, and @unbiasedscipod.

• Be a responsible reader and re-sharer of data and anecdotes. When you see a chart or statistic, check the original source and look for data from vetted sources - though the shifts at the CDC make validating sources more challenging.

• Add your own thoughts and comments when sharing content, which can help connect why you think it’s important for your network to see.

• Listen to the concerns of friends or family who have their own views and may have been influenced by some of the misinformation that abounds about vaccines. By listening, empathizing with their concerns, and then finding ways to address them, you can have a more effective conversation. I really do want to believe we’re all just trying to do the best we can for our kids.

• If you’re in the US, send a letter or call your Congresspeople to object to the firing of the ACIP members and removing recommendations for vaccinations for critical groups like children and pregnant women. The move to revisit the childhood immunization schedule in the US has concerned major medical associations, with the American Academy of Pediatrics bluntly stating that they will continue to publish their own recommended schedule for pediatricians to use “after CDC advisors sow distrust.”

In an era where anyone can set up a platform and amplify their voice, don’t underestimate the power of yours. Even if that’s in quiet, private conversations or within a parents’ group you’ve joined, speak up.

Choosing not to is a gamble for the health of our communities and our country.

Mark Palmer asked the room of data viz practitioners: do you love or loathe AI’s impact on data visualization?

Most of the hands in the room shot into the air for loathing the impacts. I’m not quiet at ‘loathe’, but I’m certainly closer to that end of the spectrum than falling in love with the new tools.

I’ve seen an increasing number of posts saying data people need to get on the genAI train or get left behind (coded for: unemployed, it seems). And while a room full of data viz people are understandingly skeptical of AI’s impact, I expect that if you asked a room of people who use data viz as a tool in their work (not their primary job) the answer may have been different.

Which left me thinking: why do I feel so hesitant to dig in and adopt these new tools?

Being an extrovert in an AI enthusiastic world

In the midst of a constant state of go, go, go over the past month between talks at the Disruptive Data Summit, Outlier, and end of school year activities with my son, the only times when I felt drained were the hours I had 1:1 with my computer.

I like spending time with people. My energy picks up and I recharge through conversation and connection. And that extends to how I like to work on data projects, which is perhaps why I’ve been so disgruntled about jumping onto the GenAI adoption wave for my work.

I don’t want AI tools to automate away my data viz design time and replace brainstorming on a white board with colleagues. I love collaborative projects where part of the magic is in the conversations and teaching clients neat tricks they can use in their data viz work. Leading workshops is one of my favorite parts of my work, where I’m in a room full of people curious about how they communicate data more effectively, asking questions and digging into their own examples.

I agree with Kathy Pham, that AI needs to pass the summer camp test before it takes over the world. Mark Zuckerberg may want us to have more AI friends to fill the companionship gaps many face, but that sounds miserable to me.

As we think about how we use new technology, part of that decision making needs to be what serves us in our work. Tools that may be an absolute delight for someone who is paralyzed by the pressure of rapid brainstorming in front of a group may not suit others, like me, quite as well.

Efficiency isn’t the only metric that matters

Joy. Delight. Fun. Beauty. Whimsy. Those all matter, in life, at work, and in artistic pursuits (which can include data visualization).

Take the common experience of meeting new people. Perhaps it would be faster to cultivate AI friends who engage and interact without the awkwardness that can come from the early stages of getting to know someone. Having moved last year away from the city I called home since 2010, I’m in my era of coffee chats and playground meetups to get to know other people in Milwaukee. There’s joy in that time being out in the world though, sitting at a coffeeshop or taking a walk down Lake Drive, and that has value too.

In data visualization design, speed to insight and efficiency aren’t the only things that matter either.

As AI tools make the creation of simple charts easier for more people, the ways we interact with data in meaningful and memorable ways will change too. One look through the shortlisted works for the Unusual category for the Information is Beautiful Awards showcases dozens of projects that often reimaging how we help people engage with data.

The winners this year include community quilting and 30 days of hand crafted visualizations from found objects. Incredible! We need people continuing to create data art, installations, and experiences that help people engage with information in unexpected spaces and formats.

GenAI tools could be useful brainstorming companions if you’re creating a data installation for the first time. But even better? If you’re designing an installation to help a community connect with data, go talk to them first. Do the user research to understand curiosities and pain points around how people make the decisions you’re looking to influence.

That applies in more corporate settings too, designing dashboards and slide decks.

Develop a deeper understanding of your audience and what they need a dashboard for before building something. You’ll ultimately save time (and money) by understanding the user needs early on rather than building something that you think is the best way to present the data. Maybe the team doesn’t even need a dashboard, and just need one number each month!

But unless those needs and pain points are written down and incorporated into the training of a custom model, you’re unlikely to get into those weeds without talking to people.

Building Trust

In our May Chart Chat, we expanded on our presentation from the Tableau Conference where we redesigned one chart four different ways.

The original chart compares data from the Edelman Trust Barometer across two years for a group of countries, focusing on both the most recent share of people who expressed trust in the media and the change from the previous period.

Andy exclusively used ClaudeAI to brainstorm, create, and iterate on his chart. Even as an expert chart creator who can translate his ideas into prompts for clear direction of the genAI tool, the tool made mistakes along the way. And frustratingly, sometimes when he would prompt Claude to make a change to resolve one issue, it would introduce new issues in the graphic.

You can see the four different makeovers below, including Andy’s co-created with Claude. I specifically wanted to humanize the data a bit more, with features like the circular waffle chart showing what share of one hundred people expressed trust in institutions based on what country and year bar was selected on the bar charts.

With the data set focused on trust, our conversation veered into bigger questions about how much we can actually trust data about trust, particularly across countries.

But there’s also a question about how much we can trust genAI tools to create accurate visualizations for us.

Andy used Claude to detect numbers and scrape them from the original graphic. But the data output had issues: Claude failed to realize the up and down arrows on the bars indicated a direction of change and a country name (India) was misplaced in the dataset, making many of the numbers incorrectly matched with their countries.

Using that data, early iterations on the chart were misleading. Fixable with a person in the driver’s seat identifying those issues, but I don’t wholly trust that everyone using charts to create graphics is doing deep due diligence of the outputs (though they should).

More broadly, as we think about the role visualizations play in connecting people with information, I’m keen to keep people leading the design process, not machines for three big reasons rooted in trust:

1. Building trust by knowing the chart creator: We know from research that one way to foster trust in (or identify where you should be skeptical of) visualizations is through knowing the creator. Particularly for sensitive or polarizing topics, where there is a lot of mis- and disinformation in the world, having a human face behind the data can help persuade when scientific data runs counter to their personal beliefs or perceptions.

2. GenAI tools make mistakes, like the data scraping errors I noted, but also hallucinate and make up information. Reports this week on internal testing data on OpenAI’s new 03 and 04 mini reasoning models reported 33%-51% hallucination rates (Forbes). Yes, people also make mistakes when visualizing data, but we’re usually at least working with real datasets.

3. Ethical issues still abound around the use of books and other copyrighted material for training large language models. For me, if I’m taking the time to buy, read, and engage with an author’s work and then draw inspiration from it, that’s the intended purpose of the book; engaging with a chat tool that leverages that author’s knowledge (often without attribution) to inform it’s replies is different. Do I trust that a genAI tool won’t make up and misattribute something to an author? Nope.

Recognize the positive disruptions, but don’t write off the human elements

I don’t loathe AI. I’ve heard from friends many anecdotes and ideas around the ways genAI tools are creating positive disruptions in our lives and in our work, and don’t want to dismiss those experiences (despite continued concerns around the environmental impacts of these technologies).

One friend uses ChatGPT to menu plan, which has reduced her mental load each week, helped to plan grocery lists, and helps her use up the produce in her crisper. A colleague uses a genAI tool to create alt text for their charts, which they then review and customize to make sure the text is accurate. Another uses Claude as a brainstorming tool for different ways to visualize data for specific scenarios.

But each of these scenarios still requires keeping a human in the loop, and focuses on a practical, specific ask.

I also recognize that genAI tools can help break down barriers to accessing and engaging with data. From asking a tool to read and interpret a chart to using AI as a co-pilot in data analysis when you don’t have deep coding capabilities, potential use cases abound. But all require that people verify the accuracy and then trust the output.

Instead of more hours brainstorming with ChatGPT, I think we need more time off of computers, connecting with data in the world, or considering novel angles and pieces of context necessary to explain why we see a pattern in data.

In one of our other Friday talks at Outlier, Obinna Iwuji and Chani Wisdom shared their curiosity around the data aurora-chasers track to find the Northern Lights, which took them on their own journey to experience this natural phenomenon. The data was a tool to get there; the feelings of awe and wonder in that experience were what mattered.

The human elements of analysis and visualization are what make this work a creative pursuit, which is why I’ll continue to bet on humans to positively disrupt how we engage with information.

I spend more time than most thinking about how to cultivate a curiosity data and answering questions with numbers for my kids, who are currently two and six years old. The seeds of being a good data communicator are sown early!

I’ve found data visualization is one of the best tools for getting kids to think more deeply about data, even if we’re not naming chart types or spinning up graphics in Tableau just yet. Obviously, charts can be colorful and fun, but they also connect with the ways our minds process information and see patterns before we even realize we’re seeing them.

In the spirit of sharing and hopefully reducing the mental load for any other moms out there hoping to raise data-curious kids, here are five ways you can help your little kids have fun with data and learn some data viz in the process.

Subscribe now

1. Be curious and answer questions with data

My kids know that my job is ‘to make pictures with numbers.’ That’s much simpler to make sense of that saying I’m a ‘data visualization designer’. Our kids don’t need jargon: they need curiosity.

I see opportunities to be curious and ask questions that we can answer with data just about everywhere, including with the objects we already have around our house. You could:

• Sort a bin of LEGO by color and stack them to see which tower is biggest.

• Separate the Halloween candy into chocolate and non-chocolate and count the pieces, comparing which number is larger (count) or which pile looks bigger (volume).

• Track how tall a kid is each month and talk about about how much they’ve grown (change over time).

We don’t need to overcomplicate things. Instead, we can model that curiosity and start sharing the questions we’re answering with data with our kids.

2. Read books that celebrate data visualization

While there are many great longer books that teach data visualization principles (like Drawing Data for Kids) well suited for older kids, I have two favorites for the littles who still love lots of pictures.

Daphne Draws Data by Cole Nussbaumer Knaflic (author of Storytelling with Data, all about data viz and communication for grownups) is a delightful, colorful romp through the world of Daphne, the data drawing dragon, whose brother doesn’t think that you can really draw data. From banana trees to outerspace, Daphne shows little readers how you can answer questions with data.

While the book has a lovely, illustrated guide to charts at the end, the real joy is that this is a beautiful kids book, which happens to encourage data curiosity in the process. Daphne isn’t just making charts at random; she’s using them to test hypotheses and solve problems. Both of my kids love this book, and we were lucky enough to have Cole do a reading and graphing activity at my son’s school!

This is Me and Only Me by Giorgia Luipi and Madeline Garner invites young readers to explore how ideas, actions, and feelings can be drawn with shape, color, position, and other encodings. The book has ample white space across the pages, where readers can draw their own iterations on the ideas, and brings concepts from the Dear Data project into a more accessible format for young ones.

All of that might sound complicated for a kids book, but it’s a beautiful way to start a kid with thinking in the language of encodings (ways to represent information) rather than the language of chart types, which can be more constraining. I’d suggest this one for first grade and up.

3. Collect and create with data to answer questions

When I teach workshops to adult learners, my suitcase might look more like I’m teaching a kids art class: post it notes, graph paper, Sharpies of all colors, pens, play doh, LEGO, sometime other random found objects…I’m all for bringing creativity and fun to rewire how we think about data and get off of the computer screen.

Last year we were at the lake and my son was obsessed with identifying and counting buoys. We also enjoyed some slow cruising around the lake, which could get a bit tedious for the kids (though did result in some great naps). Before we headed out one day, I asked which type of buoy he thought was the most common, which led to this tracker where he tallied each one he saw.

I love the drawings he created with the colors and shapes for the buoys, which made for another fun conversation around different ways we could tell buoys apart even if we couldn’t read the text in the distance. What I love about this is that the result can be a chart, but he’s also learning how to collect data.

Recognizing that buoys aren’t a topic of conversation for most little kids, set up a question that piques your little’s curiosity. Consider common logical questions like:

• Which are the biggest and the smallest? (Compare counts, weights, lengths)

• Which category has the most? (Compare counts of items or incidents)

• How long does something take? (Track and log time)

Then, find a creative way to capture the information. Maybe that’s a tracker, like our buoy counter, or a more physical way to collect information like stacking LEGO or collecting leaves in the garden and grouping by shape or color.

I’d highly recommend two things if you’re doing this kind of exercise outdoors or in another moving environment (like a car or boat): 1) use paint tape to anchor down the tracking sheet and 2) have pencils or markers that are in some kind of secured bag to avoid having them roll everywhere and result in a meltdown.

If you’re a parent with a small child, you know that’s when disaster can ensue, especially if you’re not within reach to help.

4. Experience data in creative ways

Museums are incredible places to find creative ways to engage with data. Some have dedicated sections on STEM and data, like the National Children’s Museum in Washington, DC, while others use data visualization as a tool for communicating information that even an early reader can engage with.

When we’re at an exhibit that has some kind of data component, I ask my kids about what they see (sometimes with zero interest or response back on their part, but at least I tried!) and nudge them to engage with any interactive ways to be part of plotting data.

The Feelings Cloud in D.C. (below) is a great example of using different colors of string to represent three emotions and then wrapping a string around pegs labeled with things like ‘ice cream’ and ‘bad dreams’ where the item corresponds with the feeling.

Is it precise? Absolutely not. But the goal is curiosity, not decision making.

What do we see when we look at the ‘nodes’ across the cloud? Lots of red strings around ‘getting hurt’ (upper left) and yellow around ‘friendship’ (middle right). Talk about the patterns that you see in these kinds of exhibit and ask your kids what they think. Is that how they see themselves and the world? What’s different about their experience?

5. Watch Numberblocks.

I’m sure there are other great STEM shows out there, but this list is all about what I’ve actually used in practice.

Our kids adore the Numberblocks, a BBC show where numbers are the characters and each episode is about math and patterns. For us, screentime as a tool; if it’s not something you or the littles in your life use, feel free to scroll on by this suggestion. For me, finding shows that my kids enjoy and learn from is always a win.

In Numberblocks, each number has a personality and a number of blocks equal to the number, making the math very visible. As a data viz nerd, I love the consistent use of color, shape, and block count to represent different numbers and the attention to details, like ‘3’ having three buttons and a three pointed crown and ‘5’ having a single five pointed star around one eye and giving ‘high fives.’

You can find the first season on Netflix and Amazon Prime, which introduces each number up to 20. Or, watch on their dedicated YouTubeKids channel. There are different levels to the show, so you can find everything from basic counting to complex multiplication and division.

Beware just how beloved these numbers can become. Our daughter insisted on a ‘Number 2’ theme for her birthday and our son picked Numberblock 56 for his Halloween costume. Both required quite the creative construction on my part, particularly building the wearable costume in a way that was comfortable for a 5 year old for trick or treating and for sitting down in his classroom at their Halloween party.

Over the top? Absolutely. But when your child wants to be a number for Halloween and you’re a data viz designer for your job, you say yes, because I really do believe that this young age is where a curiosity about numbers is born.

If you’re a mom, happy mother’s day. As Coco’s mom, Bella, says to Chilli in the episode Baby Race: you’re doing great.

For all of the other dads, grandparents, aunties, uncles, and friends who want to help the littles in their lives be more data curious, I hope this is a helpful list. I’d love to know what ways you help them learn to love numbers!

Sharing is caring ! Send this along to someone else who want to share the love of data with the littles in their lives.

Share

If you’re not familiar with Iron Viz, each year three Tableau designers build dashboards with a provided data set live on stage. Then, they have three minutes to tell a story with their visualizations. It’s high stakes presenting in front of 7,000+ people and always one of the highlights of the conference.

While the main event all unfolds in under an hour, conversations with contestants and past winners pointed to more than 100 hours of prep time before the conference, which is evident in the incredible dashboards they built. If you missed it, you can watch this year’s show on LinkedIn or Salesforce+.

As a judge, I had a lot of folks ask me questions about the experience judging this year’s competition. Those conversations had me reflecting on the process and challenges with judging data viz competitions, and why I still think they’re good for our industry despite valid critiques.

A glimpse into judging

I’ve had the privilege of judging many data viz competitions over the years: the Information is Beautiful Awards, Fast Company’s Innovation by Design Awards, the World Data Visualization Prize, and an Iron Viz qualifier and finals.

The judging process varies across competitions. The most common structures I’ve seen fall in one of three models:

• Individual scores, aggregated for a result: Judges receive a clear set of criteria, score the submissions, submit scores, and the totals determine who comes out on top. While this may seem the most ‘fair’ and mathematical, sometimes scores are skewed by getting judges who score submissions high or low in their ratings (which is why having multiple judges review each work is important) and sometimes a judge catches a critical issue someone else misses (like an analysis error or accessibility issue).

• Decision by committee: Judges review the submissions (typically at the shortlist phase), discuss which they would select as winners, and come to a collective decision without ever writing down any specific scores. This creates more space for conversation, but discussions can be dominated by people in the room with the strongest (or loudest) opinions.

• Individual scores first, then discussion: Judges score submissions according to scoring criteria provided by the organizers, then meet (sometimes for hours) where they get to see all of the scores and discuss who to select as the winner. Sometimes there’s a clear favorite, but other times this approach showcases the divergence in the scoring and creates space for conversation. Personally, I think this is the fairest approach, but has the challenge that it requires the most time (scoring and discussion meetings).

Judging criteria varies too. For Iron Viz, participants know they are being scored on design, analysis, and storytelling, including the stage presentation for the three finalists. For the Information is Beautiful Awards, the FAQs page outlines specific judging criteria more granularly, and there’s a matrix provided to judges to define what each score within a range means.

Listing judging criteria matters: I want every creator to put their best foot forward and get a fair assessment. When we share what we’re judging on (more specifically than ‘good design’) creators can address those criteria in their submissions. Transparency helps everyone: creators can determine what works to submit and address criteria like ‘impact’ in their cover notes, judges know exactly what they’re scoring on, and the community has visibility into the process.

Why data viz competitions and awards are good for our industry

Judging these competitions is always challenging for me because there are so many incredible projects. Objectively judging submissions created through a blend of art and science can feel like an impossible task, even with clear criteria, when you could make a case for anyone in the final stretch to take the top prize.

Some works excel in creativity and design, but the analysis is less compelling. Others have exceptionally deep analysis but have some misses in design. And how much should accessibility and other specific criteria impact the overall decision?

When we were starting up the Data Viz Society, there was heated debate in the community about the merits (and challenges) in data viz design competitions. A few of the many questions that came up included:

• Is it even possible to be objective when scoring submissions?

• How do submission fees create barriers to entry to those from under represented groups or low income countries?

• Do awards perpetuate some of the inequities in our field by elevating those with the time and flexibility to create public facing visualizations?

• Do submissions created in languages other than English face a disadvantage based on who the judges are?

• Are awards biased towards a particular type of visualization, focusing more on artistic visualizations than practical ones?

• How much does name recognition play into selection, where we continue to elevate the same people?

We talked through all of these questions (and more) when evaluating whether DVS should host a data viz awards program. Ultimately, we partnered with David MacCandless to continue the legacy of the Information is Beautiful Awards program which had paused from 2020-2021 in the midst of the pandemic, in big part because of the value to the data viz community in continuing to add to the incredible showcase of submissions.

My own thinking around the value of data viz awards was shaped by many of those conversations, with three key reasons why these competitions are valuable for our industry, particularly as AI image generation proliferates.

Let’s unpack why awards programs are good for the data viz field.

1. Inspiration

Data viz competitions create curated repositories of incredible projects. Because people submit their best projects from a year, we can go to the longlists and shortlists (not just the winners!) to discover incredible projects we may have missed in our content-saturated world.

The Entry Showcase for the Information is Beautiful Awards is one of the biggest repositories we have of data design excellence over the past decade+.

New to data visualization work and looking to explore more creative chart types? Curious about unusual mediums for presenting information? Want to dive specifically into visualization about scientific topics? It’s all right there.

You can also explore similar repositories for Malofej (data journalism) and Iron Viz (Tableau).

2. Education

I teach a data visualization foundations class for the Maryland Institute College of Art Data Analytics and Visualization MPS program. The IIB Awards showcase is one of the first places we point people to in our course, since it’s a valuable teaching tool for showcasing the breadth of works in the world of data viz.

Other competitions use a qualifier approach, providing the same dataset to all participants. Reviewing the submissions with students, you can talk through the ways the same dataset was used to travel many different analysis paths and with many different visualizations. Iron Viz uses this approach, with this year’s feeder competition using IMDB data on TV and movies.

On a very practical note, sometimes projects I point to in my classes and workshops migrate to new websites or even go offline completely. That’s happening more today than I ever would have expected, thanks to the current administration in the US and newsrooms shuttering teams like FiveThirtyEight. Having the project description and screenshots of the work on a central website, like the IIB Awards Entry Showcase, creates a lasting artifact from that work.

3. Recognition

For the individuals selected for longlists, shortlists, and as winners, adding “award winning data visualization designer” to their resume can be a game-changer in their job searches or when marketing their freelance design services. I’ve even chatted with a data journalist who got their first job because a recruiter discovered them on an IIB Awards longlist!

For our industry, competitions showcase the innovation and creativity of data visualization designers to the wider world. While it’s a joy to recognize and celebrate excellence in data visualization among peers, there’s a degree of validation (that we are an industry!) in these competitions when viewed by outsiders.

The future of design competitions

I’m curious to see what the future of design competitions will bring as tools continue to evolve, awards recognize more than just data visualization and venture into data experiences, and we see more integration of AI into data viz design workflows.

In the third installment in her series on the future of data visualization, Shirley Wu points to the Unusual category for the IIB Awards being the place she found the most inspiration, as it included projects where data was represented in sound, physical forms, books, and community installations.

Perhaps the future of our awards programs will separate and recognize different forms more explicitly. But in the meantime, dedicated awards like the Data Sonification Awards have popped up to focus more on these niche types of data experiences. (They have delightfully detailed judging criteria across their categories of Arts, Communication, and Analysis!)

My longer thoughts on AI in data visualization are more than what belong in the wrap up of this note. When considering eligibility for awards, where should the boundary be on how much AI was used in the design process?

Visualizations created wholly by prompting a genAI tool feel like they should be disqualified to me, in big part due to the ethical concerns around how underlying models were trained based on stolen intellectual property from images to books on data visualization design.

But if AI tools were used for brainstorming in the early stages of the process? Or in generating ideas for color palettes or other components? Is that just using the technology and resources available to use in ways that serve our own preferences around our design process?

This year, the IIB Awards submission form included a field for people to disclose if/how AI tools were used in the creation of the work. Like most submission information, we’re relying on self-reported information, and as AI tools get increasingly integrated into the software often used for data visualization, we may leverage those capabilities without even realizing we’re doing so which starts to muddy the “what should qualify?” waters a bit more.

Even with the big changes in how we do our work, awards programs will continue to have relevance in our industry. But for those of us privileged to judge these competitions, judging conversations will become even more challenging as we consider how to weigh how much the human part of the design process plays into selecting the winners.

The Information is Beautiful Awards 2024 shortlist was announced last week if you want to explore some of this year’s favorite works! You can also join the IIB Awards Ceremony live in Miami as part of this year’s Outlier Conference hosted by the Data Visualization Society with Alberto Cairo’s Open Visualization Academy.

The region also has phenomenal diving along coral reefs and walls, with schooling fish and an incredible biodiversity of hard and soft corals thanks to its location in the coral triangle. From massive sea fans to a dive with more than 30 sea turtles lounging along the wall of coral, it was a once-in-a-lifetime dive trip.

I don’t say once-in-a-lifetime lightly: coral reefs and sea life are being harmed by the rising ocean temperatures associated with climate change, resulting in bleaching incidents that leave reef structures barren. Whether the same spots along East Lembeh, Bangka, and the Bunaken National Marine Park will look the same if I was to return in 5 years is a real question.

In addition to the slow creep of temperatures, there was a very obvious challenge to the the ecosystem: trash.

So. Much. Plastic.

Pollution is a major issue in our oceans, so much so that there’s entire floating islands of trash collected together thanks to current movements. Where we were diving, trash would wash up on the beaches, floated on the surface, and occasionally got caught on the reef.

Each piece of plastic I saw reminded me of a quirky, memorable data viz project from a few years back by Kat Greenbook: My Plastic Footprint. She used her own trash and silicon molds to create little turtles (a thoughtful marine animal metaphor) each representing 2.7g of plastic waste.

I’ve seen a lot of scientific charts and articles about the impact of pollution on the ocean and reefs, but these little plastic turtles have stuck with me for years.

Why? They’re not overly analytical or packed with complex data. Instead, they’re personal and memorable. Even though precision and speed-to-insight gets much of the lip service around what makes a “good” visualization, memorability matters.

You can read more about Kat’s motivation on Nightingale and hear her reflect on the project in her lightning talk at Outlier 2022.

Personalizing the climate stripes

Another abstract representation of climate data is the climate stripes visualization created by Ed Hawkins, which uses color to represent temperature change aligned along an annual timeline. The design was inspired by another physical data visualization: a temperature scarf crafted by Sheldon Fiber Designs.

The stripes show annual temperature change encoded with color, so the design is slow to update with only annual changes. How do you keep people engaged with the data if change is slow to happen?

One of the powerful pieces of Kat’s plastic turtles is the personalization: this is her own data. Using the climate stripes design, the #showyourstripes website allows for a customized iteration of the climate stripes where you can see the annual temperature change for your city.

The patterns for Milwaukee, WI, where I currently live, mirror the broader global patterns with more red stripes in the most recent years, signaling alarm. But could we take personalization a step further? Instead of focusing on cities, could we focus on families?

NASA partnered with the IPCC to do just that. Climate Legacies allows you to customize a timeline with up to eight people (including name and birth year) to reflect on what temperature changes may mean for future generations.

Connecting data to impact

One challenge for the climate stripes or even the little plastic turtles is that the the numbers can seem small. How does trash counted in grams really matter? What’s the big difference with changes of 1 or 2 degrees, if you’re not as familiar with climate science?

We made a short stop in Singapore en route to Indonesia, and walked through the incredible Cloud Forest at Gardens by the Bay. As part of the exhibition, you walk through a video projection room that connects those small numbers of temperature change to real world impact with photos, video, and a voiceover describing what could happen with each additional fraction of a degree of temperature change.

The multimedia presentation is one of the most impactful I’ve seen around connecting the catastrophic impact - which we’re already seeing with more extreme weather events and disasters - to incremental changes.

You can watch the full loop from the Cloud Forest on YouTube, where another visitor kindly shared the complete video. (I would have tried to capture a full video, but had a six year old in tow who was much more interested in getting to dim sum than watching a second loop of the video.)

The takeaway

Making data memorable is necessary in advocacy work, and leveraging visually interesting, personalized forms can help achieve that goal.

But we also need to connect that data to some kind of WIIFM (or ‘what’s in it for me’) for the viewers if we want to inspire action. That doesn’t need to be with a multimedia, 20 foot tall projection video though: you can connect visualizations with photos, reflections, and stories in presentations or reports to achieve that same goal.