XXVI.2 March - April 2019
Page: 82
Digital Citation

The map is not the territory: Empathy in design

David Siegel, Susan Dray

back to top 

Design empathy is an approach that draws upon people's real-world experiences to address modern challenges. When companies allow a deep emotional understanding of people's needs to inspire them—and transform their work, their teams, and even their organization at large—they unlock the creative capacity for innovation.
    —Katja Battarbee, Jane Fulton Suri, and Suzanne Gibbs Howard from IDEO, 2014 [1]

back to top  Insights


If you ask a user experience (UX) person—academic or practitioner—whether empathy is important to design, it's hard to imagine anything other than a resounding "Yes!" Indeed, statements like the one above seem to imply that empathy is a silver bullet that will transform design and lead to innovation. Before empathy was a buzzword, many of us would still have said that helping product teams develop empathy for their users was a core function of user experience research. After all, what else could it mean to study another's experience and share those insights with others?

As often happens in business, though, once a concept like empathy catches on, it's treated like a fresh discovery. The groundswell is then translated into a small number of new techniques that instantiate the concept concretely. As these techniques become codified, too often reification takes over and their artifacts, or deliverables, seem to substitute for the more abstract virtue they supposedly represent. We act as though the map is the territory.

back to top  Mistaking the Map For the Territory

Empathy maps, originally developed and promoted by Dave Gray [2], are a hot topic that seems to be going down this path. Empathy maps seek to make visible users' underlying traits. Figure 1 shows the results of an image search for "empathy map template."

ins02.gif Figure 1. Results of an image search for "empathy map template."

Immediately apparent is the relative standardization of these maps, as if there were several essential ingredients. Of course, as with all user-centered design (UCD) and UX tools, we are encouraged to adapt empathy maps to our own specific situation. One example of how one group modified and used a template is shown in Figure 2. However, the family resemblance is still obvious.

ins03.gif Figure 2. Modified empathy-map template.

It looks to us like an orthodoxy is developing around how to construct these maps. We should be asking questions like:

  • Do these commonalities exist because there is evidence that they work? Or because putting an image in the center adds to the map's face validity (excuse the pun) as an empathy tool? Or is it because everyone went to the same workshop?
  • Are empathy maps reliable? That is, would different people representing the same user or users choose similar adjectives and arrange them in similar ways?
  • Do they meet the scientific criterion that they be, in principle, disconfirmable? What factual evidence would logically resolve a dispute about whether the depiction is true or not?
  • Do they imply design decisions?

Too many practices have been adopted without thought to these kinds of questions. Rarely do we see the kind of detailed and careful analysis of methodology that Rolf Molich has done regarding usability testing, which he reported on in the November—December 2018 issue of Interactions [3].

Some practitioners will claim we are missing the point, that the value of an empathy map is not in whether it is true or false, or tells a design team what to do. Rather, it is a stimulus for discussion and, ultimately, inspiration. But this sets a very low bar. If the validity of a map is irrelevant to its value, why does its content matter at all? If it doesn't, then we are mistaking a projective test for the team for design-relevant knowledge of the user. Increasing this risk is the fact that, as one author suggests, if a team hasn't done user research, it "can be built using knowledge from internal participants or using existing personas" [4]. Even the perception that an empathy exercise led to inspiration isn't validation—it may be a false attribution, akin to a placebo effect.

To be clear, we are not saying that the problem with tools like empathy maps is that they are flawed or have gaps; that is true of any technique. Rather, the risk is that they create an illusion of understanding if too much credence is given to the tool and the mysterious process of inspiration. Certainly, having a rough approximation as a starting point might not be a big problem if researchers remain deeply engaged in the design process on an ongoing basis—and not merely to represent their earlier findings or to do superficial usability testing without rocking the conceptual boat, but rather to frame and address new questions at a conceptual level. The risk is that treating empathy maps as more meaningful than they are will reduce teams' commitment to deep follow-up research.

back to top  Static Snapshot or Dynamic Narrative?

One issue with empathy maps and other tools like personas that capture static descriptions of users is that they do not create an understanding of the relationship among the descriptors of what drives specific behaviors, especially behaviors relevant to the design problem. Many people who use empathy maps claim that this is a feature rather than a bug: They provide a glimpse of who a user is as a whole and are not chronological or sequential. But knowing a person's characteristic thoughts and feelings does not equate to empathy. That comes from understanding a person's experiences, which are specific and dynamic. Similar experiences can be shared by people who could be described differently. By definition, empathy means we can put ourselves in each other's shoes despite our differences. This requires a narrative, not a static description.

As an example, imagine that we are working on an app to help purchasers of electronics to compare products, and one of the product categories is laptops. We do individual "shop-alongs" with several different users and, based on them, generate an empathy map we name Jessica. It includes the standard Says, Thinks, Does, and Feels quadrants. In Thinks is the entry "What am I missing?" and in Feels is "Who can I trust to help me?"


It is clear immediately that, because the different quadrants are separate, it's not possible to look at the interplay between the actions, feelings, thoughts, or expressions. When, for instance, does Jessica wonder if she might be missing something? Is it when she is checking the website, or when she observes in the store, or asks friends, or at some other time? And when does she feel unsure who to trust? The fact that each quadrant is separate makes it impossible to know. In fact, if the map is a composite, we don't know what the relationships among the elements within a quadrant are either. This makes the map both challenging to use and of questionable value as an artifact to actually help designers with their design. What design decisions come from "feeling anxious" or saying "I want something sexy"?

Knowing a person's characteristic thoughts and feelings does not equate to empathy.

A vague, general sense of knowing the user is not empathy. As Gregory Bateson said, information is a difference that makes a difference. Because designers are trying to make a difference in users' experiences, we need to be able to explain nuances of difference across those experiences. Knowing that a person feels frustrated does not tell the designer what makes the person more or less frustrated. That's the level of insight they need to support nuanced design decisions. Again, this calls for a narrative.

Some might say that we, the authors, are the ones putting too much emphasis on the artifact of the map (and on the exercises built around it), arguing that the map is simply one medium of communication among many ways to influence larger processes and that the researcher's role is to be a voice for the richer narrative behind it. But it is fair to ask if we really function that way. The pressures to reduce the concept to its representation are strong. In UX, this has happened with techniques like personas, which seem to simplify and concretize the larger effort to bring a deep understanding of the user into the development process. We all have heard about—or work in—companies that invest in the production quality of persona artifacts. Think models hired to represent the personas, glossy posters plastered in hallways, persona playing-card decks, flip books, or even life-size cutouts with persona descriptions printed on them. This emphasis on the production quality of the artifacts suggests a reification of the concept. We have to wonder if the money spent on the artifacts correlates in any way with the impact of the personas on product decisions.

Furthermore, it is not a given that narrative depth of understanding exists in the mind of the researcher who created the empathy map. Once empathy maps become part of the team culture, expectations about what these look like will develop, and researchers will feel pressure to adapt their insights to fit that format. This is likely to affect how the researcher thinks. It may be a fantasy to think they will retain the same depth of insight that they would have developed had their deliverable forced them to articulate a narrative, with its nuances of difference. By analogy, many usability researchers have adopted the practice of reporting the numbers of people in a test who experienced a particular difficulty, even though they themselves believe these numbers are often not meaningful. They justify this on the basis that their teams expect problem-frequency numbers, thinking they can ensure the numbers are not misinterpreted. But the effort of always mentioning the caveats is tiresome, and probably ineffective. Even if in principle they realize the numbers are prone to misinterpretation, when they allow it to become the norm in their organizations, it will be the norm for them as well as for their audience. Let's also keep in mind that even if the researcher is embedded in the team, the artifact will have a life of its own. The researcher will not be standing by to add the backstory every time someone refers to the map.

back to top  Empathy Shortcuts?

True empathy cannot be achieved by abstracting general user characteristics from the context of specific user experiences. That can only produce stereotyping, the opposite of the deep understanding that empathy maps supposedly create. We have nothing against the technique of shorthand descriptions of user thoughts, feelings, and actions, but, like any technique designed to contribute to a grand goal, empathy maps need to be embedded in larger processes oriented toward that goal. Taking one step in that direction, imagine using a sequence of empathy maps to summarize the user's state at different phases of a storyboard that depicts an experience holistically. There is no avoiding the need for practical techniques, but in our efforts to adapt to the pressures of the commercial world, we have to resist the tendency to reduce concepts to collections of techniques or procedures on a checklist. Empathy maps may be a useful technique, but they are certainly no silver bullet. For real empathy that can drive design, nothing substitutes for a deep and broad commitment to a full spectrum of user-centered design practices overall.

The proliferation of shortcut methods in UX practices, of which empathy maps are an example, can undermine the professional status of the field. If all that's required to produce empathy is a quick-and-dirty map summarizing impressionistic findings, and to use such maps in exercises to trigger the mysterious process of inspiration, why would we need anyone beyond a skilled generalist facilitator?

Sadly, this is something UX folks are already starting to encounter in industry. For example, usability evaluation used to have relatively high status in organizations. It now has become routinized and deskilled to the point that there are companies offering automated usability testing. There are also usability jobs that require knowledge of many applications, with usability evaluation, user research, and/or other UX-type tasks taking a backseat. And these jobs sometimes do not require even a bachelor's degree ("B.A. optional," read one ad we saw recently.) For whatever reason, perhaps because UX is no longer a novelty, or because some perceive UX's contribution as marginal, or because we ourselves have promoted the idea that quick-and-dirty techniques are enough, our profession is at risk of being relegated to the corporate basement.

back to top  References

1. Battarbee, K., Suri, J., and Howard, S. Empathy on the edge. 2014;

2. Gray, D. Updated empathy map canvas. 2017;

3. Molich, R. Are usability evaluations reproducible? Interactions 25, 6 (Nov.–Dec. 2018), 82–85.

4. Brown, J.L. Empathy mapping: A guide to getting inside a user's head. UX Booth. June 27, 2018;

back to top  Authors

David Siegel has 20 years' experience as a UX design researcher and consultant. His work has ranged from tactical usability to strategic product definition. After a four-year stint at Google, he has returned to consulting, with a focus on promoting human-centric product development practices and mentoring UX professionals.

Susan Dray has worked in HCI since before SIGCHI existed. She has been deeply involved in consulting, teaching, and mentoring. Currently VP at Large for SIGCHI, she is focusing on helping emerging HCI communities, especially in the "developing" world, and mentoring a number of future leaders.

back to top 

©2019 ACM  1072-5520/19/03  $15.00

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.

The Digital Library is published by the Association for Computing Machinery. Copyright © 2019 ACM, Inc.

Post Comment

No Comments Found