🌊 Your flow should... flow!

A few weeks ago, a reader sent me his dataviz project, asking if it could join my Dataviz Inspiration project.

It was great. And it featured something I couldn't stop looking at: an animated Sankey diagram.

It got me thinking about how we visualize flows, and where dataviz is heading on the topic.

Soooo, that's what today is about!
​

​
​
Chart types for flow

In case you've never bumped into it: I'm the author of Data To Viz, a classification of chart types built as a decision tree.

If you head to the "flow" section, you'll see a list of chart types for representing flows:

Plenty of options. Their pros and cons is not the topic today (but I strongly encourage you to explore them!).

But here's the thing: they all share the same problem.
And we can fix it.

​
​
​
The problem: where does the flow go?

Talking about flows reminds me of one of my favorite graphs ever added to the R graph gallery: a chord diagram of migration flows.

It's stunning. There's so much information packed in here.

But I have to be honest: I had to stare at it for ages to decipher it.

The main issue? Nodes are linked by chords, but it's hard to tell whether the flow goes from A to B, or from B to A.
​
​

The solution? Flow must flow.

We're in 2026. The vast majority of graphs we build are read on a screen, not on paper.

So why not make the flow... actually flow?

Which brings me back to that reader's project: a Sankey diagram where every link has little dots animating along it, mimicking the direction of the flow.

I love it. 😍

It makes the graph feel alive, and it makes one thing instantly obvious: a quantity of something is moving from one entity to another.

​
​
It used to be tough

Building this kind of graph used to be a real struggle.

First, you needed a technology customizable enough to create an animated shape like this. Namely: D3.js.

Second, you had to build the animation itself. Not a piece of cake.
​(You don't want to know how long I spent on the animated connections of my Productive R Workflow landing page 😅.)

Back when I worked at Datadog, I helped a tiny bit on the Service Map: a network diagram with flows that animate on hover. It worked beautifully and was used by thousands of customers every day to drive decisions.

​
The code, though? Phew. Hard to follow! 😅

​
A bright future?

We're living in a new dataviz era now. One where AI removes the technical complexity from our projects.

Our design eye and our inspiration should become the only things that dictate what we create!

I often talk about my AI skepticism (like last week). But there's one thing that genuinely makes me glad, and it's the reason I'm still writing about dataviz today:

I believe AI is going to push our graphs to the next level.

And I do love good graphs. 🤓

So, what do you think?

Will you try to make the flow flow next time? Or not a big fan?

See you next week,

Yan

PS: In this AI dataviz era, D3.js is the tool to learn. If you don't know where to start, check out my D3 Loves React project!
​