Exploration vs. Expression
A simple, two-pronged approach to design
I’ve been diving back into data visualization lately, and it reminded me of a fundamental fact about visualizations in general that designers, particularly digital designers, often overlook: there are two fundamental kinds of visualizations driven by two different objectives: exploration and expression. A well-designed collection of information (whether a bar chart or a web page) is expressive: it’s sent out into the world to tell a story. The story may be clear or muddled, pointed or broad, persuasive or passive, but it’s a story nonetheless. This is the part we designers tend to focus on. What am I trying to convey? What’s important to communicate? What’s secondary? What’s the tone — is the message serious or silly? What do I know about the audience and how they may interpret the message? The message may just be “here’s how to use this thing” or it may be complex, subtle, and multifaceted. But fundamentally the job of a finished information design is to be expressive, to tell a story.
The other purpose of designs is a bit more neglected. These are the working models we never show the world — the sketches, post-its, and spreadsheets we use to explore and discover possibilities. These are the lo-fi wireframes that seem to be falling out of favor these days, the makeshift drawings. And while there are certainly some times when they might be less important, they still serve an important purpose and remain an key tool in our design kit. These kind of makeshift, simplified, quick designs help us explore. They can help form a hypothesis, outline possibilities, and uncover new information. And maybe most crucially, they facilitate collaboration. For that reason alone they should get more attention. One way or another their main purpose is to explore the territory and discover new approaches.
Exploratory designs can take many forms, but the key characteristic is that they are malleable. An architect might work with cheap cardboard and masking tape to create forms. A fashion designer might drape a dress form with fabric. An artist might make small, quick sketches in a cheap sketchbook. A software designer might use a whiteboard, post-its, or a simple group of digital black and white rectangles. They’re all getting a feel for the possibilities, exploring the territory.
For digital designers, particularly those working in a startup (or startup-like environment), this exploration is often slammed by non-designers as wasteful noodling that comes up short in the all-important cost/benefit analysis. And in the wrong hands or the wrong circumstances it certainly can be a time waster. If the territory is so thoroughly explored or the end product so completely confined that there isn’t a whole lot of value in looking closer, it may not be worth the bother. Being able to see when that’s the case is an invaluable skill.
The most frustrating part, though, is that the exploration often does happen, just without any visualization and without any designers. Then designers are invited in when the exploration is complete, the story to be told is decided, and they’re left to craft an expressive visualization of the decided-upon solution. This is a mistake and hugely frustrating for most designers since our training tells us how important visualization is both to that initial exploration and in carrying the fruits of exploration forward into a final visual expression. Coming in at the end to just magic up an expressive visualization is not only much harder than it needs to be, it’s also rarely going to deliver as effective a result as when the visualization process starts during the exploration phase.
OK, maybe that’s all starting to sound a bit academic. Here’s a super simple example of what a quick exploratory visualization might look like.
Let’s say I’ve been tasked with creating an engaging and revealing infographic about the different kinds of milk (skim, 1%, 2%, whole). I start by being curious about differences between them. Why might I choose one or the other? Does 2% milk have two percent the fat content of whole milk? Or is it comprised of 2% fat and 98% other stuff? What is that other stuff? Let’s visualize some data to see what might be interesting enough to carry forward.
The words (nonfat, 1%, 2%, whole) suggest a mental model that tells a story something like the first chart [“Fat Percentage (Perceived)], while the actual data is shown below [Fat Percentage (Actual)]:
Why the dramatic difference? The words can be reasonably interpreted two ways: “2%” milk could be named for how it compares to whole milk (2% of the amount in whole milk) or it could be what percent of total weight is made up by its fat content. The word “whole” is also associated with complete, total, 100%. Of course whole milk is not made up of 100% fat, but even if we don’t consciously think it, it lurks in the background making “2%” milk sound comparatively extremely low, when in fact 2% milk has more than half (more than 60% in fact) of the fat found in whole milk, and 1% milk still has nearly a third of the fat found in whole milk. The fact that we don’t call whole milk “3.25% milk" is the root of the confusion.
So that’s an interesting story to that could be told with a visualization. If I was designing a milk infographic that might give me a good starting point, but (deadlines permitting) I’d continue playing around with visualizations of the data to see what other stories might be there to discover. How do other components (water, vitamins, other nutrients) compare to fat and across the types? What about if we measure by volume instead of weight? Or compare the amount of fat in an 8 oz glass to recommended daily limits? How might these different datasets relate to how people perceive milk’s benefits and hazards? What if we look at other dairy products (cream, butter, etc.) or compare to plant-based milks?
The charts above are not particularly well designed, just a few slight tweaks from the Excel defaults, but they’re a great first step for exploring the data and getting curious about what other stories lie within the data.
To keep this (relatively) short I’ve focused on data visualization, but the exploration/expression split can be helpful when thinking about any visual design process, and can even be used in non-visual forms like writing or the design of systems or services. The visual tools we use to explore and experiment need to be quick, malleable, and readily available. The visual tools for expression need to be more precise, controlled, and powerful to let us fully convey the story we want to tell.
That said, exploratory/expressive are more poles than separate kingdoms, and there are a few areas of overlap. Detailed prototypes, for example, can help bridge from the exploratory phase of design to the expressive phase. And a really interesting situation happens when we want to give the end users the tools to explore on their own. Deeply interactive interfaces can hand some of the power of exploration over to users, bringing things full circle, though they also exponentially up the complexity of designing the interface.
Visual tools for exploration look different from tools for expression. Exploration can be messy, simplistic, or even a bit janky. Expression leans more towards clean, controlled, and detailed. Neither one is inherently better than the other — it depends on the circumstances and the goal. Together, though, the one-two punch can lead to much bigger impact than either alone.