Resources Roundup: Using data to make people care, listen, and act
Resources for telling effective data stories, and a bad dashboard drinking game
It is a truth universally acknowledged, that a single man data scientist in possession of a good fortune dataset, must be in want of a wife jupyter notebook full of bad matplotlib plots and a captive audience to dump them on.
If you see what I did there, congratulations on being as geeky as me. And even if you didn’t, I hope you’re at least curious. With this, one of the most famous opening lines in literary history1, I want to introduce the topic of storytelling. Specifically, storytelling with data. Last resources roundup was full of concrete tips for making graphs and dashboards, and this time, I’ve got a bunch of useful tidbits for helping you make sure those graphs tell an effective data story.
In my continued pursuit for the perfect structure for these posts, I’ve added headers showing the specific topics and resources covered in each section. So feel free to jump around to find the items that most interest you, and if this structure works for you (or, you hate it), do let me know!
Wise words on the importance of data storytelling
🗂️ LinkedIn post, cool visualisation, and accompanying article.
Let’s start with why I’m sharing these resources at all: it’s because many people seem to believe data storytelling is just about making good graphs, but there’s so much more to it than that.
Brent Dykes, author of Effective Data Storytelling, put it perfectly in this LinkedIn post: he writes, “many people still view data storytelling purely from a data visualisation perspective. The result is a narrow view of what data stories are and what it takes to craft and deliver them effectively. The underlying data and narrative are equally crucial for good data stories.” Expanding on this in a full article, Brent explains that a data story is made of:
data: the essential facts underlying your presentation
narrative: no, not just the captions and axis labels, but rather, the deliberate manner in which the presentation has been structured
visuals: the means you use to present your data according to the story you have in your mind, so that others can see it too.
As he shows in the above image, you can possibly skip the visuals (for example, newspapers can quote numbers without a graph, and we still understand them), but data and narrative are essential if you want to avoid confusing or misleading your audience.
To put all this into my own words: You as the data scientist or analyst might see something interesting in a chart, but you’ll need to dig in, uncover a big picture story, and then think carefully about what to show and say (whether in spoken or written words), in order to help others see that story, too. I know, it can be tempting to just show your initial exploratory notebook, and every single graph you created during your Exploratory Data Analysis. After all, you’re probably proud of what you did, or you feel like you just don’t have time between EDA and stakeholder presentation. But this will only overwhelm your audience, and overshadow the crucial points you want to make. So repeat after me: A bunch of graphs thrown together does not a data story make.
Lessons on the many skills required of a data scientist
🗂️ Conference talk
Clearly, I fully agree with Brent that too much focus is paid to visuals, and too little to the narrative. In fact, I was thinking about this recently when I spoke at the Women in Data Science Conference. One of the speakers, Neža Štrus, made an interesting point in her talk: drawing from her own extensive experience as a lead data scientist, Neža recommended that you should pre-fill your audiences’ minds with an interpretation before you show them the data, otherwise they’ll find any meaning that suits them.
Musings on the correct order of analytics and storytelling
🗂️ Twitter thread
Does this mean the story comes first? No, I don’t think so, and I don’t think either Brent nor Neža are advocating that. I think KatGreenbrook put it excellently in this twitter thread, which is worth a read through: “It's okay to have an idea of the story before analysing the data - much like a scientist has a hypothesis before conducting an experiment. But this story idea shouldn't cause you to analyse your data differently (or worse, cherry-pick your data).“2
Frameworks for creating - and communicating - data stories
🗂️ Infographics, a book, a podcast, and a blog post
So we’ve established that analysis comes first. You’ve done exactly that, and now you have new insights you need your colleagues to know, or actions you want them to take. The data story will be your means to communicate that, and I don’t think there’s a better source of advice for how to do this than Cole Nussbaumer Knaflic, founder and CEO of Storytelling With Data. Her book (of the same name) is absolutely amazing, but don’t just take my word for it: she tweets plenty of excerpts that are already really useful on their own. Like this:
… and this one, expanding on step 6: how to tell a story:
As a wannabe novelist, I love the idea of setting up your data story in terms of a beginning (establishing the business context), a plot twist (show the problem, e.g. revenue is down, or customer complaints are up), and a resolution (in which you provide your recommendations, backed up by beautifully presented data, on what to do next). Lea Pica, of the Present Beyond Measure podcast, also takes information from this ‘hero’s journey’ structure. In this episode, she talks about using cinematic techniques to keep audiences on the edge of their seats during business presentations.
Brian Graves also provides a data storytelling framework “That Will Have Your Clients Actually Paying Attention To Your Next Data Analytics Project.” Some of the tips are a little prescriptive, but in general it’s solid, and there are some golden bits of advice too. Hit me up in the comments if you’d like me to elaborate; for now I’ll just say that ‘grouping unimportant information’ is exactly the kind of thing I need to remember, always.
Advice on making your presentations fit for purpose
🗂️ Blog post, and drinking game!
Brian’s article begins with a crucial complaint: you spend time analysing data and preparing a presentation to share it with others, but ultimately, it’s ignored. Maybe it’s because you lacked a good data story, or the solid data needed to support it. Maybe your visualisations were poorly designed, or just inappropriate. Or maybe that story was never meant to be told - or at least, not to the audience you told it to.
Ryan Dolley calls this a poor “presentation-purpose fit”, in this excellent blog post. I’ve even derived my very own bad dashboard bingo from it: Go ahead, grade one of your own presentations. For every point below that applies to you, take a shot check out the resources throughout this post, if you haven’t already.
That’s it for this resources roundup! If it helped you, check out my other posts, and subscribe for my monthly, practical updates on data science, AI, data-driven business, and more.
https://www.infoplease.com/culture-entertainment/journalism-literature/100-best-first-lines-novels
As a side note, it’s also worth checking out the data visualisations gallery of Kat’s website. I particularly like the scrolling stories section and especially the bees, and as Kat says, scrolling stories are particularly suited to telling data narratives. Also, apparently New Zealand has some kick*** data storytellers!