A ~14 min read for anyone who has ever sent a chart into the void and heard nothing back.
Let me set the scene.
It's my third week as a data intern at a mid-sized tech company. I've spent two days — two full days — building what I genuinely believed was the most beautiful dashboard of all time. Six charts. Multiple color gradients. A secondary y-axis because, honestly, it looked sophisticated. I walked into the review meeting with the energy of someone who had just invented data visualization.
The VP glanced at my screen for about four seconds.
"What am I supposed to take away from this?"
I had no answer. Because I hadn't thought about that. Not once.
That weekend, I picked up Storytelling with Data by Cole Nussbaumer Knaflic. What followed was the most humbling and genuinely useful reading experience of my early career. This isn't a book summary. This is what I actually learned — the stuff that hit me in the face and made me rethink how I communicate anything with numbers.
Let's get into it.
The Difference Between Exploring Data and Explaining Data (And Why Most People Confuse Them)
Here's the first gut-punch the book delivers: there are two completely different jobs you can do with data, and most people are doing one when they think they're doing the other.
Exploratory analysis is detective work. You're digging through data, looking for patterns, outliers, anything interesting. You're the audience. You're allowed to be messy, confused, and wrong. This is the part most of us are decent at.
Explanatory analysis is journalism. You've found the story. Now your job is to communicate it clearly to someone who doesn't have the context you do. This is the part most of us are terrible at — because we skip straight to making slides without ever deciding what story we're telling.
That review meeting I bombed? I had done great exploratory analysis. I found real patterns in the data. But I walked into that room still in detective mode, showing my evidence board instead of delivering my verdict.
The fix isn't technical. It's a mindset shift. Before you open PowerPoint, Tableau, or a Jupyter notebook to make anything presentable — stop and answer three questions:
- Who is your audience, specifically? (Not "the team." Who, exactly, will be in the room, and what do they care about?)
- What do you want them to do after seeing this? (Not "understand the data." What action, decision, or change?)
- How will you deliver it — live presentation, email, shared document?
That's it. Three questions. Answering them before touching any tool will save you more time than any productivity hack you've read.
The "Big Idea" and the 3-Minute Story: Two Things That Will Change How You Communicate Forever
Okay, two concepts from this book that I now use almost daily.
The Big Idea is a single sentence — not a bullet point, not a title, one complete sentence — that captures your entire point, why it matters, and what someone should do about it.
It sounds easy. It is not easy.
Try it right now with something you're working on. Most people produce something like: "This slide shows Q3 engagement metrics across platforms." That's a description. A Big Idea sounds like: "Mobile engagement dropped 23% in Q3 because our push notifications are sending at 3am for users in APAC, and if we fix the timezone logic we can recover it by end of Q4."
Feel the difference? One describes what you made. The other tells someone why they should care and what to do next.
The 3-minute story is the companion skill. It's your ability to explain your key message clearly, without slides, in about three minutes. Imagine someone catches you in the elevator and asks what your project found. Can you deliver the point? If not, you don't know your story well enough yet.
I used to think the slides were the presentation. The book made me realize: the slides are a visual aid for a story you should already know by heart.
Nobody Taught You How to Pick Charts, And It Shows
Raise your hand if you've ever used a pie chart in a professional setting.
(It's okay. We've all done it. This is a safe space.)
Here's the thing about chart selection — most of us pick visuals based on vibes. The data looks circular, so we pick a pie chart. There are categories, so we use a stacked bar. It looks cool, so we add a second y-axis. We never learned a framework.
Here's the simplified version of what the book teaches:
Use text when you have just one or two numbers that are genuinely important. Just write the number big. Don't hide a headline in a chart.
Use line charts for anything time-based where trend matters. This is your most powerful and underused weapon. The eye naturally follows a line and reads direction instantly.
Use bar charts for comparisons. They're simple, they work, everyone understands them. Always start your bar chart at zero — truncating the axis is one of the most common ways charts accidentally lie. And horizontal bar charts are massively underrated, especially when your category labels are long.
Use scatter plots rarely, but when you need to show a relationship between two variables — correlation, clusters, outliers — they're excellent.
Use slope charts when you're comparing two points in time across multiple categories. It's incredibly efficient. Instead of a 10-line chart mess, you get a clean "here's what changed" visual.
Avoid: pie charts (the human eye is bad at comparing angles and areas), 3D charts (they distort every number), donut charts (same problem as pie, but with a hole), and dual y-axis charts (they create confusion almost every time).
The key insight: there's rarely one "correct" chart, but there are definitely wrong ones. The right question is: what comparison or pattern am I trying to make obvious? Then pick the chart that makes that thing most obvious.
Clutter Is Actively Hurting Your Credibility
This section of the book made me feel genuinely embarrassed about past work.
There's a concept called cognitive load — basically, the mental energy your brain spends processing what it sees. When a chart is cluttered, your audience burns that energy on navigation instead of understanding. They get tired. They disengage. They ask "what am I supposed to take away from this?" — which is exactly what happened to me.
The book introduces the Gestalt principles of visual perception, which are psychological rules for how humans group things visually. The practical ones:
- Things that are close together look related. Use spacing intentionally.
- Things that are similar in color or shape look related. Be deliberate about when you make things the same color.
- Things that are enclosed together (light shading, borders) look like a group. A subtle background box can do a lot of work.
- People's eyes follow the smoothest path. Diagonal text and labels force people to rotate their heads or mentally rotate the text. Don't do it.
The decluttering checklist is almost aggressive in its simplicity:
- Remove the chart border (you don't need it)
- Remove gridlines (or make them very light grey)
- Remove data markers from line charts (unless a specific point matters)
- Clean up axis labels (fewer, simpler)
- Label data directly instead of using a legend (legends force eye travel)
- Use one consistent color palette
When I first tried this on a chart I'd made, it felt like I was deleting half my work. The result looked better in every way. The data became the star because there was nothing else competing for attention.
Your Brain Has Three Types of Memory, and They All Matter for Presentations
Quick neuroscience detour that genuinely changed how I think about making slides.
Iconic memory is your visual buffer — it captures everything you see for a fraction of a second before your brain decides what to pay attention to. It processes things like color, size, and movement before you're consciously aware of it.
Short-term (working) memory can only hold about four chunks of information at once. Four. When you put a chart with 12 data series, 3 legends, a title, axis labels, and annotations on a single slide, you are blowing past that limit immediately.
Long-term memory is where you want your key message to end up. Information gets there through repetition and through the combination of images + words (they reinforce each other).
The practical implication: simplify each slide so your audience only has to hold two or three things in working memory, and they'll remember your message.
Pre-Attentive Attributes: Directing Eyes Without Saying a Word
This is the part that feels like a superpower once you understand it.
Pre-attentive attributes are visual properties that the human eye notices before conscious attention kicks in — things like color, size, position, and contrast. In the fraction of a second before your audience "decides" to look at something, their eyes have already been directed.
The practical implication: if you make one bar orange in a chart full of grey bars, every single person in the room will look at that bar first. Before they read the title. Before they read the axis. That bar becomes the thing.
This is how you guide your audience without speaking.
The rules the book gives for color specifically:
- Use it sparingly. If everything is colorful, nothing is.
- Use it consistently. Once you pick a color to mean "this is the important thing," it should always mean that.
- Be colorblind-aware. Red-green combinations are seen differently by roughly 8% of men. Add secondary cues (line style, labels) if you have no choice.
- Be thoughtful about what colors mean culturally. Red often means danger or loss. Green means growth or positive. Don't accidentally tell a different story with your palette.
Size is the other big one. Bigger = more important. If you make something small, people will treat it as footnote material.
The "Slideument" Problem Is Real and It's Destroying Your Meetings
If you've ever made slides for a presentation and then emailed those same slides as a follow-up document, you've created what the book calls a "slideument."
The problem: presentations and documents have opposite requirements.
A presentation slide needs to be simple enough that someone can absorb it while listening to you speak. A document needs enough detail that someone can understand it without you there to explain.
Trying to do both simultaneously means you fail at both. Your presentation slides are too text-heavy to follow live. Your circulated slides are too sparse to stand alone.
The honest solution the book recommends: make two versions. Simple slides for the room. Annotated document for circulation. Yes, it's more work. But it actually works.
The shortcut most people can actually execute: for your presentation, use progressive animation to reveal data piece by piece in sync with your narration. Then, for the circulated version, send the final annotated slide with callout text explaining each key observation. Same underlying visual, different experience.
Data Has a Three-Act Structure, Just Like Every Great Story
This is where the book goes from practical to genuinely philosophical, and I mean that as a compliment.
The structure of a great play is the structure of a great presentation:
Act 1 — Setup: Establish the world. Who is the main character (the metric, the product, the team)? What was the situation? What changed?
Act 2 — Conflict: What's the problem? What happens if nothing changes? What have you tried?
Act 3 — Resolution: What's the recommended action? What does success look like?
Nancy Duarte, one of the great thinkers on presentation design, describes this as the tension between "what is" and "what could be." Your data tells the story of that gap — and your job is to make the audience care enough to close it.
Every dataset has a story in it. The mistake is presenting the data and waiting for the audience to find the story themselves. They won't. Or they'll find the wrong one. You need to have done that work beforehand and then guide them through it.
A test the book gives: if you cover up all your charts and only read your slide titles in order, does a coherent story emerge? If yes, you have horizontal logic — your titles alone tell the story. If the titles are just labels (Q1 Results, Q2 Results, Q3 Results), you have a data dump, not a presentation.
Another test: does every element on each slide directly support the message in that slide's title? If there are elements that don't — even interesting ones — cut them. This is vertical logic. One slide, one idea.
The Repetition Thing That Feels Redundant But Isn't
Here's something counterintuitive about presenting to humans: saying the same thing multiple times is not annoying. It's effective.
Working memory is fragile. By the time you're on slide 8, most people have forgotten the key point you made on slide 2. Repetition moves information from short-term to long-term memory.
The structure the book advocates:
- Tell them what you're going to tell them (opening summary)
- Tell them (the actual content)
- Tell them what you told them (closing recap)
Yes, it feels redundant to you as the presenter. You've been living with this content for weeks. But your audience is hearing it for the first time, potentially while also thinking about their next meeting. Repetition isn't condescending. It's considerate.
Actionable Takeaways (The Part You'll Actually Save)
If you've made it this far, here's the condensed playbook:
Before you build anything:
- Answer: Who? What action? How will it be delivered?
- Write your Big Idea as one complete sentence.
- Storyboard on paper before opening any tool.
When choosing your visual:
- Default to line charts for trends, bar charts for comparisons.
- When in doubt, horizontal bar chart.
- Kill the pie charts. I'm serious.
When cleaning up your visual:
- Remove every element that doesn't earn its place.
- No chart borders. Minimal gridlines. Direct labels over legends.
- Use color sparingly — one accent color max.
When presenting:
- Your slides are a visual aid for a story you already know.
- Practice your 3-minute verbal version before the meeting.
- Never read your slides aloud. Your audience is literate.
When writing your story:
- Slide titles should tell the story on their own.
- Structure: tension (what is) → conflict (why it matters) → resolution (what to do).
- Repeat your key message at the start, middle, and end.
For the long game:
- Get good at one tool (Python/Matplotlib, Tableau, even Excel properly).
- Seek feedback from people outside your team — they'll catch what you're too close to see.
- Collect examples of great data visualization and study what makes them work.
The Real Lesson (It Has Nothing to Do With Charts)
Here's what I took away that I didn't expect.
The skills in this book — understanding your audience, crafting a clear message, eliminating noise, guiding attention, telling a story with a beginning, conflict, and resolution — these are not data skills. They're communication skills. They work in presentations, in emails, in Slack messages, in job interviews.
Most early career people think being good at their craft is enough. Write clean code, run solid analyses, design polished mockups. But the people who actually advance are the ones who can make their work land — who can take something complex and make it clear to someone who doesn't have their context.
The chart is never the point. The decision the chart enables is the point.
I still think about that VP's question in my third week. "What am I supposed to take away from this?" At the time it felt like a failure. Now I treat it as the most important question in every presentation I build.
Answer that question before your audience has to ask it. That's the whole game.
If this resonated with you, I'd genuinely recommend picking up Storytelling with Data — it's a fast read and the before/after chart makeovers alone are worth it. For more in the same vein, check out eagereyes.org, flowingdata.com, and storytellingwithdata.com. Happy to discuss any of this in the comments.
Top comments (0)