Are connected scatterplots so bad?
January 31st, 2024
8 min
This article is brought to you by Datawrapper, a data visualization tool for creating charts, maps, and tables. Learn more.
Sometimes, I get emails asking me “how to get started in data visualization.” Maybe you’re one of the people wondering about that. Maybe you found a book full of data vis in a bookstore, followed somebody who makes data vis on social media, noticed that you love playing with the “charts” feature in Excel, or just really like these data visualizations that newsrooms are creating.
And now you want to create them yourself but don’t know where to get started.
This article tries to give some very practical advice on your next steps. If you ask ten people for that advice, you’ll likely get ten different answers. This is mine:
1 Find out what you like about data visualizations
2 Don’t worry too much about data or tools
3 Start to visualize data
4 Get better at visualizing data
It’s based on the experiences I’ve had and what’s worked for me. I hope some of these ideas will help you to find your own path.
If you just have a blurry feeling that “data vis is great!”, I recommend spending an afternoon figuring out what you like about data visualizations. You’ll likely want to create visualizations similar to what you already admire, so this exercise can help you find out what exactly that is:
✏️ Exercise: My data vis fanzine
Do you know roughly what you want to create? Awesome, let’s start. But before we do, let’s get some potential blockers out of the way:
Here’s what used to hold me back (and still does sometimes!) when attempting to visualize data:
It’s easy to get lost in the sea of data (“Too much to choose from…”), or to get discouraged by searching for it (“The data I want to visualize is so hard to come by!”) As everyone who’s worked on Datawrapper’s Weekly Chart blog series can tell you: You can spend hours and days searching the web for data, trying to find that one perfect CSV about your favorite topic. Don’t fall into that trap if you want to get better at data vis design. (Feel free to fall into it if you want to get better at data analysis or data research!) Your topic and data don’t need to be original. It’s totally fine to visualize what others have visualized before. Here’s how:
It’s unlikely you’ll get to work with data about that one exact thing you’ve always wanted to see visualized…because the data probably just doesn’t exist. Surprisingly many data sets, you will find, don’t exist. Or aren’t free for you to use. Or are just very well hidden on the internet. So instead of starting with a chosen topic, it’s way easier to look at what data sources exist and use them.
Simple data sources are ones that you can use right away, without much cleaning up. They’re simple CSVs and not bigger than a few megabytes. Here are some good sources for simple data:
A bit more overwhelming, but also maybe helpful: the structured archive of the Data is Plural newsletter, Storytelling with Data’s list of public data sources, and the Awesome Public Datasets collection (although the data they link to is often big — think millions of entries). Browse around until you find a data set that looks interesting, then stick with it.
Data sets can be overwhelming: So many rows and columns! Many data sets have lots of more or less interesting stories in them. You don’t need to tell them all. Try to find and visualize just one, and keep only the data that’s relevant to your story. If you still have space to show more context, you can add more data to your visualization little by little — as long as your main point is still the main point.
✏️ Exercise: What could have been?
Look at your favorite visualizations. What data did the designer probably have, but decided not to include? How did they probably simplify the data? If you want to go one step further: Find the source for the visualization and actually check what the designer left out.
This could be a subject close to your heart (a TV show you know, or data about your life), one you read a lot about in the news, or one that’s just easy to understand for most people. This way, you don’t need to read three papers and decipher five acronyms before starting to visualize.
One trap I often encounter is the fear of missing out on better data. “My data isn’t that interesting,” you might think. “There must be more interesting data out there.” There probably is — but it would also take a long time to find it. If you constantly fall for that thinking and keep looking for new data without actually visualizing it, then I recommend working with data that’s already been visualized by others. Somebody else has decided that it’s interesting, so you don’t need to doubt that anymore.
To create good visualizations, you will need to learn how to use some tools well. These could be pen and paper — but more likely will be something like Datawrapper, Observable, d3.js, ggplot2, RAWGraphs, or others. When I started with data vis, I pondered a lot about which tools to learn. R or Python? Adobe Illustrator with RAWGraphs? Am I missing out if I don’t look into Figma? What would be the easiest? What would make me a better data visualization designer?
Based on these questions, I wrote a lot about data vis tools. And as somebody who did that, let me tell you: Tools really don’t matter that much.
If you spend a year creating visualizations in Datawrapper, you’ll become a better data visualization designer. If you only use Excel during that year, the same will happen. You’ll learn about colors, phrasing, data framing, and chart types no matter which tool you’re using — and you will easily transfer these skills to other tools if needed, and they’ll stay with you longer than just the time spent using your first (or any) tool.
So just start using what you know and feel comfortable with. Or research what tools people are using who create the kind of data visualizations you want to create — and then stick with them.
With two potential blockers out of the way, let’s focus on how to actually visualize data.
You can admire all the amazing data vis out there, follow data vis designers, read all the books — but this all doesn’t make you a data visualization designer. To become a data visualization designer, you need to design data visualizations.
Well, ok, the result is more the following:
Here’s how to do that.
You’re one of those dreamers who have grand ideas but never actually sit down and work on them? Just start with 10 minutes. Sit down and tell yourself: “I will work on this data visualization for 10 minutes.” If you stop after that, you did exactly what you planned on doing. If you continue: Well, that’s also good!
Or give yourself an hour to create a visualization. Just one hour. Here’s a concrete idea:
✏️ Exercise: Visualizing “How I spent my time”
You’re disappointed with what you got after 10 minutes or an hour? Read on, we’ll get there.
“I’ll create one small chart a day/week.” “Every Monday night, I sit down and try to create a data vis.” “I’ll create 20 new charts in the next year”. Routines and goals like this can help you to actually visualize data, and making it public can increase the pressure (in a good way!). No wonder that quite a few people have set themselves routines with a clear goal or end:
There are also community initiatives that you can try. Because everybody is visualizing a certain data set or working on the same prompt, it’s fun to compare results. And many people are happy to give feedback:
Good data vis design is made of so, so many little details that you probably don’t pay attention to when looking at it. That’s why copying great data visualizations is one of the best ways to get better yourself.
✏️ Exercise: Copycat
Not convinced by this exercise? UI/UX designer Erik Kennedy makes a great case for copywork in the interface design field in his article “Copy If You Can: Improving Your UI Design Skills With Copywork” — a lot of this is applicable to data visualizations too.
A similar exercise with more creative freedom is to design a new visualization with your own topic and data, but copying the design of a visualization you like. Ask yourself: “What would the creator of this visualization do? What design choices would they make? What vis type, fonts, colors, title, annotations would they use?”
This exercise goes even further: You don’t have a visualization as a reference anymore, but just an art or graphic design style (here are nine classic and 12 new ones). Browse Pinterest (possible starting point), Instagram (possible starting point), or collections like but does it float and Designspiration until you find something that makes you think, “Hey, let’s try to translate that into a data vis.” Then try it.
All of these little exercises should be fun. You might need a few minutes to get into the flow, but you should get there eventually. If they’re not fun after 30 minutes, stop doing them. Data visualization should be a satisfying activity, and you want to finish feeling like “this was time well spent,” not “urgh.”
This is sorted by how effective I believe these exercises are at getting you better at data visualization:
Have I mentioned that, to get better at data vis, you’ll need to create more data vis? I think I have, but in case I haven’t, let me just state that again: To get better at data vis, you’ll need to create more data vis. It is truly the most effective way I know to get better at something: Just keep doing it.
Not every chart or map you build will turn out awesome. Even if you build an awesome data visualization one day — the next one might not be awesome again.
Personally, that dynamic held me back quite a bit over the years. Every time I created a data visualization I was really proud of, I was either disappointed the next time I visualized data (”that looks only meh compared to last time!”), or I didn’t dare to do so in the first place (”it won’t turn out that great again.”)
But keep going. Keep creating, and keep sharing. Because over time, the average quality of your visualizations will become better:
Don’t compare your latest visualization with one from last week. Compare it with one you created six or twelve months ago.
Seeing yourself improve can be a huge boost in motivation to keep going. And to improve more quickly, it’s crucial to know what’s not working in your visualization and how to make it better — either by asking others, or by training your critical eye.
One of the biggest reasons I wanted to work at Datawrapper in 2017 was that Gregor Aisch was working there. And Gregor, I knew, is an excellent chart designer. I believed that whatever visualizations I created, he would be able to improve them. And he did. I would create a chart, put my best effort into it, then show it to Gregor. And Gregor would make it so much better, every single time. I might have learned more about data visualizations by watching him (and others) improve my charts than by any other method.
If you have a great data visualization designer sitting next to you, make use of them: Show them your work, and ask what they would improve. If you don’t happen to share desks with such a person, consider applying to jobs or internships where you will.
Not an option, either? You can always email data vis professionals for feedback, too:
They might give you feedback during their free time, so the more concrete your feedback questions are, the better. Not: “Can you please give me feedback on this chart?” (If that’s what you want, consider paying the person you’re asking: ”What’s your hourly rate? Can I pay for in-depth feedback?”) It’s better to ask for the part you’re the least sure about, like:
Consider not asking the designers with tens of thousands of followers, busy jobs, teaching/speaker side gigs, and book deals. You’re more likely to get replies from somebody who gets an email asking for feedback once a year instead of once a week.
✏️ Exercise: Ask an expert
Asking strangers feels weird to you? Consider going to local meetups or international conferences to get to know people before you email them a few days later: “Hi X, it was great meeting you the other day! (I was the one with the dinosaur sweater who couldn’t figure out how to use that fancy bottle opener.) I was wondering if you have five minutes to give me feedback on…”
Some organizations out there do amazing work for people who want to start or get better at visualizing data. Joining a community can be a great way to keep you accountable and get inspired… and, yes, you can ask them for feedback:
Whatever community you join: Give, not just get. It’s more likely that people will help you out when you help them somehow too. And please don’t spam with three posts about your work, but post one and see how it goes.
✏️ Exercise: Ask the community
You’ll improve at visualizing data by understanding what could be improved about a visualization you created. But it’s not necessary to always ask people for feedback to get there. It’s also helpful to train yourself to look at your work like you’ve never seen it before. To do so, here are two ideas:
✏️ Exercise: Fresh eyes
Experiment with how long to set aside your visualization (two hours? Two days? Two weeks?), and with what you’re doing while you wait. Do you find more things to improve after looking at lots of other great data visualizations in between?
✏️ Exercise: Compare with the pros
By training your critical eye on visualizations from others, you’ll get better at turning it on your own visualizations too.
✏️ Exercise: What to improve?
Optionally, think about how the chart was designed. What were the likely constraints? How does the visualization deal with them?
While getting inspiration alone won’t make you a better data vis designer, it can help to motivate you: Every time I come back from a great conference, I get at least three ideas for new data visualizations I could create and can’t wait to do so. Here’s what works best for me:
Also useful to get inspired:
If you’re reading this with a bit more experience, I wonder: What worked well for you? Let me and future readers know in the comments below, or write to me at lisa@datawrapper.de. I’m looking forward to hearing from you.
Comments