We stand with Ukraine and our team members from Ukraine. Here are ways you can help

Metaphorical Analysis

by Cyd Harrell, Jodi Leo
13 min read
Share this post on


Playful exercises with metaphor can draw new insights out of piles of research data.

“There is an interesting dance here, in that data (observations, equations, structural formulas, spectra) are useless without the narrative, theoretical framework to make a story out of them.”—Raold Hoffman

Every UX designer faced with a 6-inch stack of research notes and a looming deadline has wanted to take a nap and wake up with the most important insights neatly tagged. While we can’t offer that exactly, there is an incredibly powerful and fun shortcut that many designers aren’t using: working with metaphor.

You might have learned about metaphor as a literary device, or a component of interface design (e.g. the folder and file system as an organizational concept); it is those, but more importantly it’s a highly sophisticated form of human play that allows us to find associations and hypotheses quickly through the power of unconscious processing. The technique we’ll lay out here involves imagining your challenge playing out in domains outside your own, and seeing what happens when you attempt to make new combinations. The approach assumes that you’ve begun synthesizing your verbatim notes (or raw data) by highlighting surprises over trends, taking special notice of their deep motivation to use your product/service.


“Have you guys heard any of my metaphors yet? Well come on, sit on grandpa’s lap as I tell you how infections are criminals; immune system’s the police.”—Dr. Gregory House (House, television series)

It is important to distinguish that we are not talking about design motifs like skeumorphs, or casually invoked rhetorical devices that compress an idea for the sake of convenience, as in the two metaphors above from Dr. House in which “infections are criminals” and the “immune system is the police.”

The metaphors we’re talking about come from an unmasked and vulnerable state of mind that we don’t often associate with the workplace, where we have to be buttoned up and professional, so as not to appear too flaky. Yet this is when you get to be truly creative, oh creative person! The state from which metaphor can act as guide is aptly described by innovation strategist, Tihamér von Ghyczy, as “engulfed in a curious mixture of meaning and incomprehension. It is in such a delicately unsettled state of mind that we are most open to creative ways of looking at things.” He distinguishes this type as conceptual metaphor from the Dr. House-like rhetorical kind, which ties it well to the type of free association needed to make the shortcut work.

Generating a strong metaphor for a user experience provides a testable statement very efficiently; this isn’t hard to do, but it doesn’t happen through the activities that we traditionally think of as data analysis. Buckling down and categorizing faster, or harder, actually doesn’t work as well as taking a playful swim in a qualitative data set and letting your mind wander through it.

Look at metaphor in its native environment for a minute: poetry. What makes good poems work? Let’s take a stanza from Theodore Roethke’s I Knew a Woman:

Let seed be grass and grass turn into hay:
I’m martyr to a motion not my own;
What’s freedom for? To know eternity.
I swear she cast a shadow white as stone.

Why does a line like “I swear she cast a shadow white as stone” (which is complete nonsense out of context) make so much sense? The line shortens the distance between stone, shadow, the color white, and the act of oath-making in a way that perfectly demonstrates infatuation. It’s not as much fun unpacked, is it? Try again: The poet is saying the woman holds sway over him to such an extent that he fully believes things in the world like shadows and stones to be the opposite color from what they are, and to have a close connection despite being at opposite ends of the scale of substantiality. You can start to imagine what the data analysis version might look like.

Substantial ←————–> Ethereal
Black ←————-> White
Sane ←————-> Raving
Serious (sworn) ←———> Flip

The presence of the lovely woman is a variable that correlates with significant changes in the poet’s perceptions along all four axes.

We all have a chart like this in a research analysis somewhere (possibly even in a presentation). And it communicates much less, and with much less economy, than “I swear she cast a shadow white as stone.”

So why are we using traditional techniques instead of shortcutting with metaphors? The reality is that traditional data analysis is an easily justifiable thing for UX teams to spend project time on. Numbers seem solid, real, important, bellwethers to conversion. Often those descriptions are as true as they seem, but there are many things numbers and hard analysis won’t do for us. And when we are neck deep in data, there’s a “can’t see the forest for the trees” problem. It can be hard to get a wide enough angle on our questions to draw broad inferences. Metaphor solves this problem by laughing at it.

Wide Angle View

“Some of the most creative breakthroughs occur when an idea that works well in one domain is transplanted in another.”—Mihaly Csikszentmihalyi

If you’re used to approaching data analysis from a quantification angle, your metaphorical muscles are probably a bit rusty. It helps to think of metaphor as akin to sketching: a skill that comes more easily to some than others, but that all of us can learn. Much like with sketching, it helps to avoid being self-critical. Your weirdest ideas may be your best, and even metaphors that clunk can help you find your way to better ones. On the bright side, it’s also like a paper snowflake: just make a couple of cuts for the beauty to reveal itself.

The classic, and classically ridiculous “if bachelor number three were a car…” question isn’t a crazy place to start. Test drive the experiment by playing it with your workmates as bachelors—it forces you to create and defend associations between things that have nothing to do with each other, which is the heart of the game. Once you have an interesting association, don’t stop. The most fruitful metaphors will carry through multiple iterations and across multiple aspects of a problem.

A strong metaphor is extensible. It doesn’t just give you a little giggle and that’s it. You can keep pushing it and thinking up aspects. If bachelor three is really a BMW, we can predict a lot of things about him. We can come up with what kind of phone he’d have (cough, iPhone in a fancy case, cough), and what he’d order at the bar, how he’d behave in a line, and if we have data about his behaviors, we can test whether that metaphor is truly apt. It gives us a lot of side characteristics and components to look at.

As you begin your exploration, set your sights on quantity over quality and come ready with a few domains that extend outside of the one you’re working in (think of it as mental oscillation between two or more domains). Be as silly as you like: while an elves/dwarves/men metaphor probably won’t explain everything in your analysis, it could easily yield useful insights for some groupings and it will certainly put your mind into play mode.

Choose an instigator on your team who will be good at helping you widen the field of association and is prone to experiencing the world in novel or original ways. The leader will provoke ideation and push the team in an extreme direction, listening for wherever a person says, “it’s like…” or “it’s almost like…”

When “it’s like” is uttered, the instigator will ask:

  • “What does this remind us of?”
  • “What kind of relationship? Are they dating?”
  • “Is it formal or casual?”
  • “What animal would they be?”

When thinking about the product under scrutiny and its participants, triptychs can also work, like:

  • Elf, dwarf, or man?
  • Date, marry, or dump?
  • Fast food, farmers market, or fine dining?
  • Detective novel, self-help book, or bible?
  • Seeker, student, or guru?

Next, take a walk. Our everyday environment relies heavily on metaphor. Imagine yourself in a different scene and force yourself to stop taking what you see at face value—shift the metaphors around. What are some scenes you can use?

  • Desert: stop sign could be a cactus (vertical part of landscape) could be the sensation of discovering that that water up ahead is a mirage
  • Spaceship
  • Underwater
  • Disaster zone
  • Mission control
  • Riding a beam of light (like Albert Einstein did when he was 16, which eventually led him to his theory of special relativity!)

All of this will feel silly and cheesy and maybe like you shouldn’t be billing hours. Eventually you’ll check the insights gleaned from metaphor games against your reams of behavioral data, and then use them to generate rich metrics to further validate the findings. But for now, allow yourself to play. Humor and emotion are sources of deep insight about the world. Metaphors that appeal to our mammalian or even reptilian instincts can be especially productive: food and sex, family and status, belonging, yearning; fight or flight.

It’s true that most of the time we’re not making products related to fight or flight or social belonging. But within what we capture when we interview users, we can extend what we find toward extremes. If you’ve interviewed 10 bargain hunters and discovered a strong theme that they enjoy the game of finding the cheapest price even at a cost of their time, what does that tell you? It could say simply “there’s a strong competitive motivation.” That’s very hard to imagine as a metaphor. Maybe another Roethke could pull it off, but your average poet would be stumped. Extend the strong competitive motivation to its extreme, however, and your thought process will pass through “winning” (more interesting) and reach “victory” or “triumph”—concepts that are much more inspiring to the imagination.

From there, what is another kind of person that seeks triumph? Does gladiator resonate? Or wrestler? (Here you will have to listen to your own mind and pick one to start with.) Suppose it’s a wrestler. What else do we know about wrestling? We know it’s a mental game that requires a ton of preparation and a stock of clever tricks. We know it’s one-on-one, with a clear focus on a single opponent. We know people sometimes do obsessive and self-destructive things in pursuit of this kind of triumph. We can then look back at the data collected from the bargain hunters to see if there are analogs for any of these behaviors. (Hm. Hadn’t we recorded that these people pursue winning even if it costs them more time than the savings are worth?) If there aren’t, we can move on right away. But if there are, we’ve just provided a really useful key for how to parse that 6-inch stack of data.

A party invitation is a wonderful and broadly applicable metaphor generation game. If the product you’re working on was a party, who would be invited? What would the dress code be? Would there be food? How about an open bar? Could invitees bring guests? If you create a metaphorical invitation that resonates with your team, you can evaluate your initial experience against the elements of the invitation.

One of our favorite metaphors from Bolt Peters: a reader’s desires in news analogized to desire for food. Did they want a fine restaurant (everything prepared and presented to them by an expert), a farmer’s market (the very best ingredients laid out for them to pick and choose), fast food (not worried about quality as long as it’s there when they happen to need it), a CSA box (something they know they should do, but assembled by someone else so they don’t have to think about it). While we generated these metaphors from responses to open-ended questions, they turned out to correlate closely with sets of concrete behaviors observed during field research, including visit frequency, start pages, and methods of navigating the site: behaviors that were very important to the client.

Once you have sets of useful metaphors, you’ll have a wealth of design inspiration. But how do you sell it to clients used to numbers? A good metaphor generates values that can be the basis of metrics and can dictate your proportions and statistical baselines. Going back to our wrestling/bargain-hunting example, an experience that delivers triumph will cause certain observable behaviors. Measuring those will both validate the metaphor and provide a much more useful metric to assess the success of the design.

When Should You Use This Method?

Doing metaphorical analysis (at least openly) works best for clients who see their goal as taking the right risks with their designs as opposed to reducing the risk of design decisions. It is, in fact, very effective for the latter, but play is often deprecated in risk-averse cultures and it may make such clients uncomfortable. On the same news-based analysis where we successfully developed the food metaphor, we were searching for a way to represent a clear rift in the user sample between people who looked to news to converge their thinking around particular issues and those who looked to it to spark new threads of thought.

Stating it in that way wasn’t working. We ran through many possible metaphors: feathers and rocks, arrows, nozzles, gates versus open fields. We kept returning to feathers and rocks, but feeling that it didn’t resonate. Eventually a team member said, “Does this work better in a water context?” And after everyone laughed, we shifted the feather and rock to an anchor and a sail. Suddenly we had something that explained a lot. Both an anchor and a sail are desirable things on a boat, for very different reasons: the sail offers the possibility of going anywhere, somewhat at your own control, but somewhat dependent on the prevailing winds and currents. An anchor offers certainty and safety—a way to make it all stop.

We brought the anchor/sail metaphor to our client as a distilled extreme to describe the motivation in each of the personas that emerged from our research. They come to … get anchored for the day, get grounded and sated, to be surprised, to skim the headlines, to escape their local “bubble.” Bearing in mind that the goal of the research was to create archetypes to evoke real empathy and serve as a launchpad for future design efforts, we hoped that this type of active characteristic would inspire design.

As we got closer to the final analysis, the client began to look ardently for correlations to quantitative traffic analysis. It’s almost as thought the same impulse that pushes to let numbers dictate the direction of one’s product also seeks to invalidate qualitative richness or authenticity. The real human stuff can be messy, even when it is distilled into an intrinsic set of guides like “anchor” and “sail.” It can be tough for even the pluckiest client to defend an alternate domain, like sailing or desire for food, in a board meeting about subscriber demographics. Set clear expectations that your output will convey the transfer of emotions, which is what metaphor has done for centuries.

How to Set Yourself Up for Success

You can practice your metaphor generation skills, and we hope you will. One more thing you’ll need in order to do this well is to record the right kind of data in your user research so that you’ll have a rich field to work in. If you are taking verbatim notes or making transcripts, you’re well on your way, but there are a few extra things you can do. First off, make sure you’re scripting to capture emotion: ask questions that will allow space for emotional responses, and follow interviewing practices that will put the participant at ease.

Don’t ask participants any metaphorical questions directly. Instead, it can often help to create a score sheet based around behavioral and emotional dimensions. By defining two sides of a continuum so you can score which side users land on, dimensions help you group behaviors and motivations effectively. These dimensions are hard to get right, so we’ve adopted the practice of generating a lot of hypothetical dimensions and assuming a third to half of them will turn out to be irrelevant within the first few interviews. Interesting dimensions are things like “latest vs. best” for a concept of quality, or “habit vs. ritual” for a daily behavior.

Other examples:

  • Serendipity vs. reliability (to characterize the participant’s desire for product predictability)
  • Own vs. borrow (likelihood to share socially)
  • Struggle vs. self-guided (way-finding patterns)
  • Rich vs. lean (level of interaction the person has with the product)
  • Perfect vs. good enough (expectation of the site)
  • Content meets them vs. destination (modes of consuming of product content)

Again, these aren’t things to ask the participant directly, just something to score in the background. They provide a good field for checking your metaphors but can also be useful to generate them if you’re stuck. If you find clusters along the dimensions, can you come up with a metaphor that embodies all the characteristics from a cluster? That’s a useful exercise as well.


“We do not choose our metaphors, our metaphors choose us.”—Allen Kurzwell

The primary goal of sharing this work is to show you some of the shortcuts we’ve found using metaphor to power through daunting amounts of raw data quickly. There has also been a quiet call in our community for an approach to rival data-centrism in product development. Edward Tufte, in his article, “Metaphors, analogies, thought mappings,” said, “a good way for contributors to develop this thread would be to provide examples of specific metaphors that work in the sense that they have some explanatory power.” We hope our specific examples provide an inspiring way forward that lends itself to more precision and authenticity.

As always, use your powers for good. We experienced a bit of zeitgeist as we were typing the final words in this article, when our colleague, Frank Chimero, blogged about metaphor. The title of his post says it all: “The Cloud is Heavy and Design Isn’t Invisible.” He goes on to caution that, “a metaphor can clarify or obscure. The most dangerous ones do both. They illuminate one characteristic of a concept, but also throw another (usually less favorable) aspect into shadow.” Tufte has a similar example of Defense Secretary, Donald Rumsfeld “comparing insurgent attacks to a ‘fruit bowl’—a metaphor that stripped them of all urgency and emotion,” but let’s not even go there.

If you get lost in your exploration or feel you are headed down a path of obfuscation, take it back to the roots of the word, “metaphor:” “phor” meaning “to carry or bear,” “meta,” meaning. This transfer of meaning is both like glass in its capacity for transparency and mirror in how it reflects back on the design in progress.


Image of anchor courtesy Shutterstock.

post authorCyd Harrell

Cyd Harrell,

Cyd Harrell is UX Evangelist at Code for America, where she gets to spend all day helping create better experiences for citizens. She encourages like-minded UXers to apply for the 2014 CfA Fellowship. Before its acquisition in June 2012, she was VP of UX Research at Bolt Peters in San Francisco. 

post authorJodi Leo

Jodi Leo,

Jodi is a designer and champion of the arts. In 2000, she founded the Grange Open House for the Arts, aiming to give artists, designers and musicians an open space to build and sustain their practices. Since 2000, 14 artists have worked and flourished there with residencies resulting in 3 published books, 8 released recordings, 12 art exhibitions, several grants and successful creative ventures.

Currently, she is VP of Design at CapTap, which uses technology and data to connect America’s small businesses quickly to the working capital that they need to grow and succeed. At CapTap, she oversees a group of designers, researchers, and content strategists. Jodi is also an instructor at the School of Visual Arts MFA in Interaction Design, where she teaches Research Methods.

Prior to CapTap and SVA, she was a director of design at agencies Barbarian Group and Bolt | Peters, where she worked with brands like Apple, CNN, GE, Google, Lexus, Mozilla, The New York Times, and VMware, to iterate quickly and extend products into new markets and platforms. 

Before her time at agencies, she was director of design and innovation at Fidelity Investments, where she simplified dozens of complex online transactions for 20+ million 401(k) investors into a single product that resulted in a patent and tripled participation rates within the first 12-months of launch. 

In addition to patents, Jodi has received a half dozen industry awards. Most recently, she was awarded a 2013 People's Choice Award by the Interaction Design Association as a member of the UX4Good team. She mentors teams at TechStars NY and Lean Startup Machine.


Related Articles

Article by Eleanor Hecks
8 Key Metrics to Measure and Analyze in UX Research
  • The article outlines eight essential metrics for effective UX research, ranging from time on page to social media saturation
  • The author emphasizes the significance of these metrics in enhancing user experience and boosting brand growth.

Share:8 Key Metrics to Measure and Analyze in UX Research
6 min read

Repetitiveness, complicated setups, and lack of personalization deter users.

Article by Marlynn Wei
​6 Ways to Improve Psychological AI Apps and Chatbots
  • Personalized feedback, high-quality dynamic conversations, and a streamlined setup improve user engagement.
  • People dislike an overly scripted and repetitive AI chatbot that bottlenecks access to other features.
  • Tracking is a feature that engages users and develops an “observer mind,” enhancing awareness and change.
  • New research shows that users are less engaged in AI apps and chatbots that are repetitive, lack personalized advice, and have long or glitchy setup processes.
Share:​6 Ways to Improve Psychological AI Apps and Chatbots
3 min read

As consumers’ privacy concerns continue to grow, so should our attention to addressing privacy issues as user experience designers.

Article by Robert Stribley
Designing for Privacy in an Increasingly Public World
  • The article delves into the rising importance of addressing privacy concerns in user experience design, offering insights and best practices for designers and emphasizing the role of client cooperation in safeguarding user privacy.
Share:Designing for Privacy in an Increasingly Public World
9 min read

Did you know UX Magazine hosts the most popular podcast about conversational AI?

Listen to Invisible Machines

This website uses cookies to ensure you get the best experience on our website. Check our privacy policy and