Good Charts: The HBR Guide to Making Smarter, More Persuasive Data Visualizations1
Visualization is an abstraction, a way to reduce complexity,
Seven “visual variables” with which we “encode” data: position, size, shape, color, brightness, orientation, and texture.
The principle of expressiveness: Say everything you want to say—no more, no less—and don’t mislead.
The principle of effectiveness: Use the best method available for showing your data.
Research shows that charts help people see and correct their factual misperceptions when they’re uncertain or lack strong opinions about a topic. But when we understand a topic well or feel deep opposition to the idea being presented, visuals don’t persuade us.
Charts that present ideas counter to our strongly held beliefs threaten our sense of identity; when that happens, simply presenting more and more visuals to prove a point seems to backfire. (The research goes on to suggest that what’s more persuasive in those situations is affirmation—being reminded that we’re good, thoughtful people.)
Representation is sometimes a more intuitive and human way to understand values than statistics
We do math with our eyes
Communicate ideas, not data sets.
[When reading a chart] We don’t go in order.
A chart reader may not get to the title at the top until well after she has started scanning the visual middle. She may jump around.
We go where our eyes are stimulated to go.
Although the challenges of producing good visual communication—to achieve clarity, focus, and simplicity—are in some ways no different from those of producing any other communication, they’re in other ways distinct and more difficult.
We see first what stands out. Change and difference—peaks.
Despite their position, titles aren’t usually the first thing a chart reader sees. Rather, they’re clues to help us find the meaning that started to emerge when we looked at the picture.
We see only a few things at once.
A good guide is that with more than five to ten variables or elements, individual meaning begins to fade.
Bad complexity neither elucidates important salient points nor shows coherent broader trends. It will obfuscate, frustrate, and ultimate convey trendlessness and confusion to the viewer.
Good complexity, in contrast, emerges from visualizations that use more data than humans can reasonably process to form a few salient points.
We seek meaning and make connections.
We process visual information thousands of times more efficiently than we do text.
The ability to find meaning so efficiently may be a blessing in a fire, but it can also lead us to construct false narratives from data visualizations.
We can’t help making connections in what we’re presented with. Anything that stands out becomes part of the narrative we’re trying to form,
Good visual communication should be used not just to produce better answers but also to generate better conversations.
We rely on conventions and metaphors.
Time visualizations can move in any spatial direction and remain factually accurate. But we’ve learned to think of it as moving left to right on a page or a screen, and back to forward in three-dimensional space.
Conventions are a form of expectation, and our brains use experience and expectation as cognitive shortcuts so that we don’t have to process everything anew every time we see it.
Up is good, down is bad. North is up, south is down. Researchers have found that we even connect those metaphors to value judgments. For example, because south is “down,” we think it’s easier to go in that direction than to go north, which requires us to go “up.” There are others: Red is negative, green positive. But red sometimes means “hot” or “active” (which can be thought of as positive), and in those cases, blue means “cold” or “inactive.” Hierarchies move from the top down. Lighter color shades are “emptier” or lower than darker ones.
Like colors mean like items—the
Color saturation indicates higher and lower values—lighter
Categories are arranged and plotted from one extreme to another—we
Like colors mean like items.
Color saturation indicates a progression of values.
Categories are arranged and plotted from one extreme to another.
If something is hard to perceive, people not only struggle to find the right meaning, but judge it less favorably.
It’s not the chart that they’ll judge harshly if the meaning is hard to find; it’s the information itself. They’ll consider it less credible.
If you don’t understand these basic tenets of how we see information—if your charts don’t make what’s important stand out; if complex data doesn’t coalesce into a few clear ideas; if the information visualized fosters a false narrative; if unconventional visual techniques confuse your viewers—then you’ve promised music but delivered noise.
Visuals aren’t read in a predictable, linear way, as text is. Instead, we look first at the visual and then scan the chart for contextual clues about what is important.
Our eyes go directly to change and difference, such as unique colors, steep curves, clusters, or outliers.
We seek meaning and make connections.
If visual elements are presented together, they should be related in a meaningful way; otherwise, viewers will construct false narratives
A good way to start thinking visually is to consider two questions about the nature and purpose of your visualization: . Is the information conceptual or data-driven? . Am I declaring something or exploring something?
Either you’re visualizing concepts and qualitative information or you’re plotting data and information.
There are three broad categories of purpose—declarative, confirmatory, and exploratory—the second two of which are related.
Managers most often work with declarative visualizations.
Declarative viz shouldn’t preclude conversation about the idea presented; a good one will generate discussion.
But let’s say you think but you’re not sure. Now your purpose is confirmatory, and you’ll dip into the same data to create visuals that will show whether or not your hypothesis holds.
[In such cases, your visuals] don’t always have to be presentation-worthy.
If your hypothesis is confirmed, it may well lead to a declarative visualization you present to the boss.
Or maybe you don’t know what you’re looking for. Instead, you want to mine this workbook to see what patterns, trends, and anomalies emerge.
This is exploratory work—rougher still in design, usually iterative, sometimes interactive.
It’s a kind of data brainstorming that can deliver insights.
Other ways to ask the purpose question: “Do I need to give the answers, to check my answers, or to look for answers?” Or “Am I presenting ideas, researching ideas, or seeking ideas?”
declarative, conceptual visualizations simplify complex ideas by drawing on people’s ability to understand metaphors (trees, bridges) and simple conventions (circles, hierarchies). Org charts, decision trees, and cycle diagrams are classic examples of idea illustration.
Because the discipline and boundaries of data sets aren’t built in to idea illustration, they must be self-imposed.
The skills required here are similar to what a text editor brings to a manuscript, channeling the creative impulse into the clearest, simplest thing.
Visual discovery. This is the most complicated category, because in truth it’s actually two categories.
A hypothesis can’t be confirmed or disproved without data.
Confirmation is a kind of focused exploration, whereas true exploration is more open-ended.
What am I working on? • What am I trying to say or show (or prove or learn)? • Why?
What am I trying to show or say (or learn, or prove)?
Just as rough drafts improve even staff memos and other prosaic writing, sketches will make even simple charts better.
Line graphs are usually a good starting point for trends.
clarity can be achieved by removing nonessential information.
Each element is unique and supports the visual.
lack clarity because elements are used to describe the chart’s structure rather than support the idea being conveyed.
Supporting elements that have a finer purpose—that augment rather than just repeat—enhance clarity.
Describe the chart’s idea rather than its structure.
Ambiguity forces us to stop, refocus, and think about the visual rather than the idea.
But simple isn’t always clear, and clear doesn’t have to be simple.
relative simplicity—how little you can show and still convey your idea clearly.
obvious path to simplicity is to remove unnecessary things from the chart, leaving only what’s valuable to communicating your message.
It’s also hard to edit yourself. If you didn’t think some element was necessary, you probably wouldn’t have included it.
A chart presented on paper or on a personal screen—a format in which viewers can spend time with the visual—may benefit from more detail that allows the viewer to reference individual values and explore
a chart broadcast in a presentation—when you want the audience to understand the visual in seconds—fewer structural elements will reduce distractions and make it easier to focus on the broad ideas.
Are we meant to focus on the specific values, or on the overall shape of the thing we’re looking at?
If you feel that it’s necessary to show every value, a table may be a better option:
There’s no right answer here without knowing the context.
In general, a design will feel simpler if you apply as few unique attributes as possible.
The more color differences they see, the more they have to work to figure out what the distinctions represent.
This deep-seated belief that more is better, that complex equals smart, must be eradicated. That’s not what makes charts good.
You’re trying to reveal truths dormant in the data;
A play-by-play announcer calls the action, describing mostly what’s actually happening on the field; a color commentator influences our sense of the game’s narrative.
What often makes a chart persuasive is how easily you draw people’s attention to the main idea so that they can process it.
If you make an idea easy to access, viewers will often find it more appealing and persuasive.
Words that describe statistical trends (increasing, declining, underserved) naturally give way to words that describe feelings (hurting, helping, hungry).
Make it stand out.
Emphasize and isolate your main idea.
The most obvious and common form of emphasis is color.
When every variable gets a bright color; no one variable stands out.
Once it’s time to talk, discuss the idea, not the object that shows the idea.
Notice how discussing ideas instead of explaining the data and structure naturally leads to more human-centered language.
Stories increase empathy, understanding, and recall. Storytelling is persuasive.
our brains grab on to narrative, and we need it to make sense of statistics. And narrative emerges much more quickly when it’s visual.
Braess’s paradox, a principle of traffic management developed by the mathematician Dietrich Braess, which states that adding route options (new roads, new lanes) to congested roadways can decrease traffic performance.
It has been applied to phenomena other than traffic, including power transmission (performance declined after systems were decentralized), protection of endangered species (the prospects for many species improve when one species goes extinct), and crowd control (multiple paths from a concourse to a seat in an arena make it take longer to get to seats).