Product teams, including our own, often interpret empathy as evidence. However, in practice, empathy is actually something that drives us to seek evidence. By observing and evaluating various examples within Shopify, we have identified 4 traps that are common in the way empathy is manifested. We modelled the relationship between empathy, problems, evidence, and decisions to provide strategies for how to use empathy effectively while being sympathetic to its limitations. Since empathy drives us to seek evidence, and thus cannot be considered evidence itself, empathy must be used at an appropriate level of abstraction throughout the product decision-making process in order to influence good decisions.
INTRODUCTION
Intentionally gaining empathy for users has become an essential way of developing better products or services. Empathy, or getting as close as possible to users to understand their needs, is now a prolific term in the field of user experience (UX). It has become a constant theme in books, articles, conferences, events, career postings, and is a core concept in user-centred design.
At Shopify, empathy is at the very core of our business. As a commerce platform for entrepreneurs (we refer to them as ‘merchants’), Shopify’s mission is to ‘make commerce better for everyone’. We have an organisational culture team dedicated to inspiring us to care about our users and being ‘merchant obsessed’ is encoded as one of our internal company values. The idea of empathy is ever-present in our work at Shopify—we even specify it as a required skill in our recruitment of UX practitioners. Yet in UX, we have not critically explored the relationship between empathy and the way we make decisions about the products we build and the people for whom we build them. Through the research presented in this paper, we uncover how empathy has influenced Shopify’s approach to product development and decision-making.
Empathy is subjective and context-dependent
Currently, in UX, empathy has been positioned as a singular characteristic: we aim to ‘have empathy for users’. Crucial as it may be in motivating us to solve user problems, developing this kind of emotional connection is a complex process. People experience and feel empathy in different ways depending on the individual or the subject that elicits the empathetic responses. Empathy, therefore, creates a paradox. While we use it to design for people who are not necessarily like us, there are various influences at play that can make us experience empathy differently depending on the context. This effect could, therefore, make empathy a problematic tool for making decisions if it is used in a way that is not sympathetic to its potential to create bias.
Empathy is a process involving two parts. Firstly, cognitive empathy allows us to understand another person’s context, emotions, goals, and motivations. Secondly, emotional or affective empathy is the ability to respond to those things with the appropriate emotions (Hodges and Myers, 2007; Esser, 2018). So empathy is rooted in subjective experience. Research shows that people generally feel more empathy for those closest to, and most like, themselves (Arbuckle et al., 2012; Bloom, 2016; Gutsell et al., 2010).
Our capacity for empathy can also be affected by scale. In other words, there is an adverse correlation between the size of a group and the extent to which we are empathetic towards them. (Slovic, 2007; Cameron, 2017). Cameron and Paine (2011) suggest that this is because empathy can be emotionally exhausting. Even when we engage with it in small doses, we expect the needs of large groups to be potentially overwhelming. This makes us regulate our emotions to prevent these feelings, thus diluting the effects of empathy. As a result, empathy can become more potent with individuals and magnifies when we can personally relate to them.
Differences in how we experience empathy can also depend on factors such as cognitive conditions (Decety & Moriguchi, 2007; Baron-Cohen et al., 2015; Bora et al., 2008; Tone & Tully, 2014) and genetics. Warrier et al. (2018) showed that our predisposition to be empathetic can be influenced by our biology. They found that as much as 10% of our empathetic potential can be determined by genetics. This study also showed that on average, women are more empathetic than men. This is believed to be based on non-genetic biological factors such as socialization.
UX research, empathy, and evidence
UX researchers have the potential to succumb to the influences of empathy in how we make decisions. We have the skills to control for bias when representing information. We adopt research methods to avoid skewing results, and where we cannot control for that, we report on the impact it could have. But researchers are not immune to the bias that can come from developing empathy for users. When we recruit participants into research, we see a name and a profile and we start to paint a picture. It is what our brains are best at: filling in the gaps. This increases when we connect with the person, have something in common and empathize with their story. Due to this connection, researchers could overemphasize the hardships of a certain user they relate to, wanting to solve their problems over those of other users.
UX researchers also have the responsibility for evangelizing and developing user empathy among stakeholders and teams. When teams care deeply about users, they are better equipped to build the right solutions, in the right way. UX research helps to bridge the gap between product teams and users by generating empathy strategically. Researchers hold an incredible amount of responsibility in how we collect, analyse, and disseminate user information to product teams. We leverage empathy by annotating our insights with things like video clips, photos, quotes, and storytelling. This creates emotive responses, which help teams care about users. We should be considerate of this responsibility and the way in which we facilitate empathy within teams and the impact that can have on how teams make decisions on which problems to solve. For this paper, we are going to be talking about how we make decisions, specifically with regard to solving user problems, and the role empathy plays in this process.
As a product team, we don’t just use our level of emotional connection to users to decide how to act on our research, we also look to prove a problem is true before choosing to solve it. This is because time wasted on solving the wrong problem, or only symptoms of a problem, rather than the root cause can result in wasted resources. To establish a problem as ‘true’ we need evidence that accurately measures its impact from valid and reliable information. But like empathy, evidence is a complex concept. In a general sense, it is comprised of facts or information that indicates whether a belief or proposition is true. This makes evidence appear easy to judge, as it uses facts to define truth. However, this is complicated by views from prominent theorists that challenge the possibility for objectivity when interpretation is inherently biased. Donna Haraway (1991), for example, claims that all knowledge is ‘situated’ and that no perspective can be prefered over another. Our view of the world can cloud how we interpret and understand information, which impacts whether we believe something to be true or not.
This tells us that evidence can be taken out of context depending on the person interpreting it. In order to view evidence objectively, it must be tightly coupled with its original goal and the context in which the information was gathered. We see this in journalism often: a blindly called-out percentage or misused quote to reinforce a point. These are weak forms of evidence because they are deliberately taken out of context to influence belief about a subject in the desired way. This means that evidence is also problematic for decision making if not used responsibly.
Ultimately, an awareness of where knowledge comes from reinforces our approach to UX research. For one, demanding that evidence is robust in validating a problem means that relevant information must come from studies that rely on observable behaviours to bolster the conviction that further research is unlikely to change the degree of confidence we have in our findings (Hodgson, 2017). Furthermore, during evaluation, we must be aware of the ways in which our biases can limit the positive impact of our research. The very nature of the abductive reasoning process we use to develop products makes it unlikely we will have complete data informing our decisions when we gather evidence. As a result, we make decisions as we go, based on our current information, to constantly assess our level of certainty in the evidence we have. Empathy presents a particularly challenging component of this assessment because it can bias the way in which we interpret information as evidence and therefore our notion of what is important.
Since we can never be fully aware of a user’s lived experience, recognizing bias means acknowledging that this can be a limitation of our research. A study carried out by Dan Ariely and colleagues (2012) demonstrates empathy and decision making in practice and informs our approach in understanding commerce experiences. It shows the difference between judgments made based on your own experience versus observing that of another. In this research, Ariely and colleagues studied the IKEA effect with origami. Builders made origami creations while buyers observed them. When complete, both were asked how much they would pay for the creation. Across the sample, buyers were willing to pay 5 times less than what builders would pay themselves. The study indicates that people assign more value to the things they put time into and build themselves. It is also an example of the differences in experience between employees at Shopify who observe merchants creating businesses, versus the entrepreneurs who are building and creating these businesses. We often believe that exposing ourselves to the lives of these users will enable us to make good decisions for them, but this study shows that this is a potential trap. Being an observer is not the same as being the user. We did not build their business ourselves and we are unlikely to make the same decisions for it as they would for themselves.
EMPATHY AT SHOPIFY
Today over 600,000 businesses in 175 countries use the Shopify platform. While many entrepreneurs seek the same goal of having a successful business, there is an increasing amount of differences in their needs, contexts, and motivations in pursuit of that goal. This diversity makes our mission statement ‘make commerce better for everyone’ a much bigger challenge for Shopify. As Shopify has scaled, grown, and gained success, we’ve begun to view this challenge as an opportunity to innovatively reimagine the commerce landscape and meet our mission statement in new ways.
With scale comes an abundance of information, and parsing all that information to make decisions can be challenging. When you add the effect empathy can have on our judgments, this could be problematic in making good decisions. This is especially noticeable when we don’t use empathy responsibly. By admitting that we will never have adequate evidence to make objective decisions, we are actually getting closer to responding to our users’ real needs. It can be argued that since we are so embedded in our particular contexts, all knowledge is a product of our environment. However, by questioning the methods by which we seek evidence, we can get closer to a more rigorous approach to creating better products. In this research paper, we look at the effect of empathy on decision making and discuss how to further develop the UX industry definition of how to use it in a more responsible way.
At Shopify, empathy is an ever-present theme. It is encoded in our company values, and there are ample opportunities for employees to develop it for the entrepreneurs who use the platform. It is positioned as a way to help us make good product decisions for our users by prioritising the achievement of their success as a way to ensure our own.
Examples of how developing empathy is encouraged at Shopify:
- New hires design and build a new online commerce store to better understand how the platform works and how it supports entrepreneurship.
- Employees can shadow support calls to hear firsthand the challenges and pain points of users.
- All employees are encouraged to regularly do on-site visits (i.e. in a Shopify user’s retail location) to observe how they run their business and use Shopify.
- Communications and company-wide events always feature stories from users and the businesses they run.
- Quarterly internal hackathons are inspired by and revolve around solving particular user problems that are not currently being addressed.
- Common frustrations coming from users via customer support get automatically posted to team Slack channels.
- Team members of all disciplines are encouraged to observe and take notes during usability tests, validation, and interviews.
Further, Shopify embeds researchers and data scientists within product teams to use their expertise to ensure learnings and user insights are shared more readily. Research is integral to understanding needs and contexts, as well as ensuring products are developed in a way that solves real problems. As researchers, we focus on both generative and feature development research, depending on team needs and priorities. We are enabled to choose the best methodologies based on our research questions. We bring the team closer to the users they design for, rather than being a bottleneck to information. Everyone is involved in research and builds empathy through direct exposure to our users.
The opportunities for teams to build empathy for users come in many forms so that teams effectively understand user needs and their environments. However, as our user base grows, so too do the needs and contexts to support. Building understanding of our users is a constant necessity that will never be completely fulfilled. Empathy is something we continuously develop and we’ve created an ideal environment to facilitate this, but it’s currently focused on gaining empathy generally and not necessarily how to apply it.
METHOD
Our method for investigating empathy and the ways it is applied in decision making at Shopify involved interviewing employees from across the organisation. We selected 13 participants based on their varying tenure and exposure to different projects. We also explored viewpoints from different disciplines—namely UX, Engineering, Data Science, Human Relations, and Culture—to assess the impact of empathy from 2 different angles: how we apply it to building products and how we instil it as a company ethos.
We interviewed each person one-on-one for 45 minutes about their experiences in product teams and how empathy had played a role. Those interviews were crafted around the following research questions:
- What are the definitions of empathy and evidence based on different individuals experiences and disciplines?
- What is the relationship between empathy, evidence, and decision making?
- How does empathy inform and influence their day-to-day?
- What are teams’ decision-making processes in how they approach solving user problems? What impact does empathy have on this?
- What types of projects or initiatives cause empathy to be used successfully or unsuccessfully in decision making?
We also observed how and when our own team used empathy and evidence. Specifically, we reviewed project documentation, sat in project meetings, team conversations, discussions with stakeholders, show-and-tells, and UX team rituals. We wrote down the information that was used to make decisions or inform direction and how that information was presented. We conducted observations over a period of 6 weeks. For the first 4 weeks, we observed teams not connected to our own products in order to control for any bias we would have in being decision makers in the group. We collated the data and came to some thematic groupings for how empathy impacts decision making. We then spent the final 2 weeks using this as a framework for evaluating within our own team. One example of a UX team ritual we observed was called ‘Angry Thursdays’. It is a half hour per week focused on user struggles. During this session, one of the UX researchers highlights user frustrations from recent research and/or support calls. The goal is for the team to regularly stay close to and maintain empathy for users. We observed how information was disseminated to the team and how the team used that to make decisions.
In addition, we documented when empathy was used at the company level as a way to engage and motivate employees. We observed broader UX events, such as Shopify’s internal UX conference. Empathy was the subject of numerous talks. One talk’s key message was that empathy is a secret superpower and will allow you to have exponential impact through your work. At the company level, we looked for use of the term empathy, a focus on highlighting our users in some way, and the method by which the information was presented.
From all our observations and interviews, we collated the findings and analysed them by impact and effectiveness on decision-making. We noted the ways that empathy had not been sympathetic to its potential bias based on our original research, such as empathy being treated as a singular characteristic. This brought us to 4 thematic traps.
FINDINGS
Product decision-making process
Our findings uncovered a structure in the way in which product teams, in theory, make decisions on which problems to solve and how. We modelled this process to contextualise our findings for the impact of empathy. The product decision-making process has three stages: problem, evidence, and decision. We found empathy traps manifest in some or all of these stages.
Figure 1. Product decision-making process
Working out which problems to solve and how to prioritize is a complex process that has to take into account many different factors. The foundation of decision-making is the problem itself. The ‘problem’ here refers to a hypothesis we have about what we are trying to solve, which is based on any of the information we have access to. It is a hypothesis because we don’t necessarily know the details of the problem and therefore whether it is true. Before we can make good decisions, we must validate the problem through evidence.
The second stage of the process is evidence. Information becomes evidence when it answers our question or validates the hypothesis about a problem. At Shopify, evidence could be derived from qualitative or quantitative information from research, data science, and/or customer support, but not exclusively from these sources. Evidence can come from information that is either already collected, or it can be gathered and generated within projects. If the information contradicts the hypothesis, this likely means that, for the current belief about the problem, we don’t have enough information, the problem is different than we expected, or it turns out not to be a problem at all.
The top stage of the process is the decision. The decision relates to what we will do about the problem now that we have evidence to validate that it is true and assess its impact. While this process is not an exhaustive account of decision-making, it is important to remember that, in addition to evidence; opinions, expertise, business goals, and technical constraints should come into consideration in making decisions about a problem.
4 empathy traps in decision making
Our success in building empathy at Shopify has brought us a unique challenge. While empathy’s ubiquity in how product teams approach solving problems is encouraging, through our research we uncovered 4 traps that can happen in decision-making processes when empathy has been commodified: creating fake empathy, unbalanced use of empathy, using empathy to force decisions, and superficial empathy for show. For each, we have visualised the trap’s problematic relationship between empathy and the decision-making process to highlight where it presents challenges.
Trap 1: Creating fake empathy – Creating fake empathy is a trap in how empathy manifests itself in the problem stage of decision making. This trap is prevalent in the problem stage when someone experiences a problem through created experience and then applies the experience as an understanding of the user’s perspective. But because an individual experience is fundamentally different from person to person, each individual will have a unique perspective on the same situation. Ignoring this trap could mislead our understanding of a problem or misidentify a problem where none actually exists.
Figure 2. Trap 1: Creating fake empathy
One example of this trap was observed through an initiative encouraging employees to be entrepreneurial and start their own businesses using Shopify. As is common with companies of all kinds, Shopify wants its employees to ‘dog-food’ the product. This is the practice of using the product in order to experience how it works. While not being exclusively targeted at building empathy, according to our participants it was an impactful by-product. But employees have different information and motivations than the entrepreneurs who use our platform.
“Although well-meaning, it misses the mark. The stakes aren’t real when you have a steady paycheck coming in, when you have access to more information or resources than the general public, and there are no real consequences if your business doesn’t succeed. That’s not really realistic because the context is so different.” Data Scientist
Unlike our user base, Shopify employees are not relying on their business to support their livelihoods. Employees also have extra context from knowing the platform deeply and that information impacts their approach to using it and how they go about starting a business.
“Someone on my team runs a store and will use themselves as the archetype of who we’re building for. Problem is, they have so much tech and product context that users who don’t work at Shopify don’t, that they are a bad template for users. Their context is not the same.” Developer
This program was an incredibly positive exercise for building product understanding along with identifying the kinds of challenges users might be up against. But it does not make us entrepreneurs in the same way our users are. In fact, we found the pain points that employees experienced were often spotlighted, creating a bias for seeing and assigning more weight when those things arose in research and/or customer support data.
“The idea of needing to put ourselves in the shoes of users is not the best illustration of what empathy is. Kids playing with aluminum foil pretending to be an astronaut doesn’t make them understand what it’s like to actually be an astronaut. Pretending to be entrepreneurs is projecting our own biases in that scenario. Fake empathy makes you make bad decisions [based] on assumptions and stereotypical emotions.” Design Lead
Context is an important part of an experience causing emotions to be subjective. Projecting our own experiences as a true reflection of those of our users will incur inaccuracies from this subjectivity, rendering the ‘empathy’ we feel fake, as it’s based on a personal and biased view. We could miss a key detail or highlight something as important to us that isn’t as important to our users. There is not enough information from these pretend scenarios alone to translate those feelings into good product decisions.
Creating fake empathy conflates information from one’s own experience with the experience of actual users, mistaking it for a true representation of a user problem. Misguided decisions can come from the belief that, because the problem experienced is the same on the surface, the feelings are also the same, and therefore judgments made will be equivalent.
Trap 2: Unbalanced use of empathy – Unbalanced use of empathy happens people put more weight on information that elicits a strong empathetic response over other pieces of information. This trap causes every step of the product decision-making process to be overshadowed by, and optimized for, the experience of select users or more salient parts of their experience. It is the challenge when encoding the information too quickly goes from experiencing cognitive empathy (understanding) to emotional empathy (assigning emotion to that understanding). Emotional empathy clouds judgements by spotlighting the experience due to a bias created by increased emotion. It tricks people into believing some information is more important than other pieces of information.
One participant in our study sample gave an example of this trap that they experienced when their team was developing a filters feature. In research, one user had demonstrated frustration in not knowing where to begin in selecting and applying filters. The team ended up overusing this qualitative account in deciding how to develop the feature, without considering the size or nuances of this problem throughout the population. In subsequent testing with a larger group, they found the proposed solution actually created friction for most users. When making the decision about what to solve and why, the team misinterpreted the problem by putting too much emphasis on the empathetic impact of one story and optimized for removing a negative experience of one individual.
“People will take anecdotal evidence and assume it’s the full picture. We’re only starting to learn to build user empathy, but not necessarily what to do with it.” Developer
Figure 3. Trap 2: Unbalanced use of empathy
Another example was when a product team made the decision to redesign the workflow of a key section of the product. The solution at the time worked well for a specific segment of users, but it did not scale to the majority of use cases. The team used empathy for that smaller group and decided not to take action for a length of time. The decision to redesign the functionality eventually came from prioritising the needs of the larger group, even though it would be a significant adjustment for the users it currently served well.
Unbalanced use of empathy can also show up deeper in the details of a problem when people rely on salient parts of information without enough context. This happens when teams select certain aspects of information and make decisions based on a misunderstanding of its significance. People have more empathy for the parts of the experience on the fringes because they stand out and more easily create an emotive response. People, therefore, find it harder to connect with issues that fall in the middle, which are more nuanced and difficult to categorise clearly. This overemphasis on salience can lead to acting on things with a biased understanding.
One participant gave us a project example related to an older version of the first page users see when they log in to Shopify (a content feed of information related to a user’s business). Shallow understanding of significant business milestones, such as getting to a first sale, meant celebratory messages were created in abundance and shown when such milestones were reached. This was problematic because, despite the milestones being achievements, the messages themselves lacked the appropriate language to recognize that reaching them can also be incredibly stressful. These milestones did not always signify celebration in the minds of users. These messages were based on a conclusion drawn from the salience of highly empathetic pieces of information related to those milestones. While this page still celebrates these moments with users, due to gathering a deeper understanding of these experiences from further pieces of research, as a company we have adjusted our tone to relate more to the context of the milestone.
“We almost glorify empathy as the way to represent our own assumptions of a merchant’s success without understanding the full perspective or considering all the factors. We built a part of our product as a celebration screen, highlighting how much revenue a merchant has made. But what if it’s zero dollars all the time and they’ve put their life savings into this?” Design Lead
We also observed this misuse of empathy in our internal communication channels. Shopify is an open and collaborative company, so any employee can highlight a problem with our product to the relevant team. Along with this, customer support can ask product team members real-time questions coming directly from users via these channels to get help finding answers and troubleshooting. Although it is a great way to provide support and keep teams close to user problems, it also creates salience in the team’s mind because they know this is a real user struggling with this problem right now. As a result, there is a prolonged effect of wanting to prioritize and solve symptomatic problems over those that are systemic, and sometimes more impactful.
“I’ve definitely seen people amplify a human frustration they’ve witnessed, even if the data says otherwise. And I’ve also seen projects that experiment data said should be shut down persist because there were a handful of stories about how users ‘really like it.’” UX Researcher
Spotlighting specific experiences based on empathy leads to information being insufficiently framed by its context. This can cause short-term misguided decisions. Due to the level of empathy salient information can create, teams are more motivated to fix these problems. However, that motivation should be channelled into gathering context and deeply understanding the complexity of the problem to be solved. From our examples, we observed the impact of empathy in this trap was to actually cause lower empathy overall because decisions were often lacking a true understanding of users’ full experiences.
We found that this trap was less prevalent for participants with longer tenure. They tended to be less reactive to this kind of information and wanted to seek further context by building up a deep understanding of the complexities of running a business.
Trap 3: Using empathy to force decisions – We observed individuals leveraging empathy in team decision-making scenarios, such as team project alignment sessions and reviews, to influence one decision over another. It is important to note that this is not a malicious activity, rather, these team members cared and wanted to improve the platform for users. This trap builds on the previous trap, unbalanced use of empathy, as it is often those singular, more salient experiences being used as evidence and causing too much influence at the point of making a decision. This trap can actively stunt other individuals’ contributions and cause important viewpoints and information to be left out.
Figure 4. Trap 3: Using empathy to force decisions
“In the absence of enough information, everyone says things like ‘as a user, I would do x’ to get what they want, or talk about one user’s experience to represent most of our user base, which actually shows a lack of empathy.” Developer
An example of this can occur after employees do an on-site visit to learn about a user’s business and how they use Shopify. Employees will often come back from these unique visits deeply motivated by empathy to address a specific problem they just witnessed. They recount their observations to the team to inspire more empathy and motivate a desire for action. This pressure can also manifest itself later in prioritising problems, as references back to these visits to influence decisions. But the information obtained should be understood and treated as general findings that broaden the team’s worldview or as a starting point for additional research, not as evidence on which to base decisions.
To complicate things even more, if a team member argues against information that is underpinned by empathy, especially when it relates to the success or failure of a user’s business, that person can appear unfeeling. In seeking to avoid this, individuals feel discouraged from contributing to a conversation where empathy is being used to force or influence a decision. This happens even though everyone wants the same outcome: building the best product experience and making good decisions on how to get there. However, if individuals fail to contribute other viewpoints, this potentially important information will be left out of the decision-making process.
“Empathy is a good way to shine a light on what you know nothing about, makes you aware of what you don’t know, and can debunk our assumptions about situational use cases. Empathy is good for awareness, but definitely not for decision making.” Developer
Another example of this trap is when employees have decided on an idea and implicitly use it as a frame of reference for seeking evidence to support it. This behaviour has been observed when colleagues reached out to UX researchers for specific user quotes and other insights at the final push to get things started or committed to. The information requested is often more emotionally influential, rather than broad and quantitative.
Empathy should not be a singular driving force of decision making. Using empathy to force decisions flips the product decision-making process upside down to be about the decision coming first. As a result, many of the opportunities to understand the problem, along with gathering the evidence required to support good decisions, can be missed.
Trap 4: Superficial empathy for show – Superficial empathy for show happens when it is popularised without a deep understanding of how to gain and apply it. Having empathy becomes a catch-all concept that you can check off once you have ‘done empathy’. There is good intention behind this, but it’s based on a simplified definition of gaining empathy. It can quickly make empathy a vanity metric for ‘good product development’. Teams end up doing it because they should care, not because they necessarily do care. Empathy becomes the act of doing ‘research’ and doing ‘research’ in its practice alone is seen as ‘having empathy’.
Figure 5. Trap 4: Superficial empathy for show
“Empathy can be used as a checklist, which is not authentic. People can grab a couple of things and think ‘yes, I did the empathy part of this’. But what if you grab the wrong items? Empathy, after all, is not about finding one use case. It’s about taking cues from multiple experiences.” Developer
As an example, we often heard things like, ‘Have you spoken to a user about this?’ after conclusions had already been reached. This act suggested that having empathy was in seeking a user’s seal of approval and was enough evidence to proceed with a decision. This way of speaking about research as a proxy for having empathy made it a separate entity and something that was disconnected from the process leading to decisions.
We also saw examples of this trap through confirmation bias in how people sought evidence: making a decision and then validating it with evidence after the fact. This is seeking a problem for a solution and has too much bias in how the information is filtered and framed.
“Empathy is a term that’s thrown around. We’ve done testing and talked to users, so we have empathy, but it almost feels like we’re already working towards that. Is that empathy if we’re already heavily leaning towards a decision and not looking at things critically?” Design Lead
The impact of this trap on decision making is that it isolates empathy as an item on a checklist to complete separately. Using empathy as an amendment to a decision already made creates bias in how we seek and interpret information, which misleads teams to conclude that information is evidence when it isn’t.
Summary of empathy traps
The challenge with all these traps is not the intent to seek and have empathy, but how we interpret the feeling, assign it value, and translate it into decisions. It can be intense enough that it becomes evidence in people’s minds, creates weight, and biases decisions. It then inhibits careful interpretation and objective balancing of all the information available.
Here is a summary of the four empathy traps we identified:
- 1) Creating fake empathy by believing we feel the same as our users when engaging in empathy-building exercises. It happens when we transfer our experiences as accurate representations of the experiences our users feel. Direct decisions on this information lack context and are skewed to our beliefs.
- 2) Unbalanced use of empathy by over-indexing on it, like spotlighting specific experiences and assigning more weight to some information over others. It happens when we use shallow but salient information based on the empathy it evokes. This can cause short-term and misguided decisions without enough context.
- 3) Using empathy to force decision-making can stunt others’ contributions and cause the omission of other important perspectives. People leverage empathy to elicit an emotional response in others to sway the outcome of decision making in a specific direction.
- 4) Superficial empathy for show is a simplistic interpretation of empathy as an item on a checklist. It isolates empathy and causes it to be an afterthought. Decisions are often already made and empathy becomes a confirmation bias activity to justify that decision.
In UX at Shopify, empathy is about caring for users and wanting to develop solutions to enable them to run a successful business. This is a big responsibility which, in and of itself, can evoke emotion among our teams. Shopify is dealing with entrepreneurs’ livelihoods and therefore the decisions made can have serious consequences. Good decisions are made from synthesizing various pieces of information and viewpoints. Pieces of information that create empathy are valuable, but cannot be used in isolation.
DISCUSSION
Discovering these 4 traps is not to discount the role of empathy in how we work, it is a powerful tool in getting teams to care about users and is a vital first step toward attentive action.
Empathy is not evidence, it is a driver to seek evidence. – In industry settings, empathy can successfully impact a variety of product decisions, starting as an effective motivator for teams to care about users. Throughout our examples, it was this motivation that drove action. But channelling this motivation is crucial. Rather than acting on our emotions, empathy is more beneficial if we harness the drive it creates in order to learn more about our users and their needs.
Empathy provides a strong signal, revealing the need for more information and a deeper context. With empathy for our users, it can give us a glimpse into their point of view when we do not have much information. This works on the basis that we do not trick ourselves into believing that shallow information is enough, or that empathy is an item on a checklist and separate from our responsibilities.
This signal often presents itself as a moral feeling. When someone presents an idea that you feel uneasy or unsure about but don’t necessarily know why, that is likely empathy at play, giving you an implicit sense that something doesn’t feel right. Use this feeling to go and gather more information, and be empathetic to the fact that we might not have enough understanding yet.
For empathy to act as a signal to gather more context, we first need to build a deep understanding of our users. We saw in our sample that it was the participants who had been with the company the longest who had this sense, but they also pointed out that we cannot get complacent and think we know everything. This is a continuous process to keep gaining empathy and adapting the way we approach using empathy, especially as we scale.
Empathy Decision Model
In order to understand the right relationship between empathy and decisions, we created the Empathy Decision Model. The mapped relationship in this model shows that empathy is a layer on top of each stage of the decision-making process, with more influence at the bottom, and less as you move to the top. As a layer, it is also no longer isolated as an afterthought but is instead a requisite part of each stage.
The main role of empathy is that it is a signal to seek more information and helps us be critical about what we believe to be true. At the problem stage, it should be highly influential to encourage us to dig further into what we know about a problem, our users, and their context. It should hold a mirror up to our understanding and make us critically reflect on any assumptions we might have and/or our lack of information. It drives a conscientious approach that stops us from pretending we know and moves us to an informed place where we know as much as possible. Drawing assumptions about our users from not enough information can incur poor decisions, which is ultimately not an empathetic act.
Figure 5. Empathy Decision Model
At the evidence stage, it provides a critical lens through which to evaluate the information we have as evidence of the problem, helping us determine if it reliably represents aspects of the problem or our users’ context related to that problem. The best way to do this is to review information through a research question or hypothesis as a counter to any potential bias you might have in what you look for and interpret as evidence.
When making a decision, empathy is less directly influential. At this stage, it should have been contextualised, rationalised, and carefully considered in the evidence we have. It is key we no longer go directly from empathy to decisions. For good decisions, we need to have empathy, but at a level of abstraction. Empathy never goes away, so when decisions are made, it is implicit because it has been baked in throughout the process.
Empathy at the decision stage can also play an explicit role as a final check in decision making. It likely shows up as a doubtful feeling about a decision. This happens when empathy from your deep understanding triggers your brain to tell you to stop, take a step back, and review what we know; that we should seek more information and critically evaluate the evidence. Ultimately, having 100% perfect information, and therefore evidence, is improbable. Most of the time decisions come from the best information we have at the time, which is weighed against other details like technical constraints, business goals, effort, time, and resources.
Weighing a decision based on risk is a critical step. We assess risk based on the potential impact of a decision and how confident we are with the information we have. Some decisions are riskier than others depending on how much of a negative impact the decision could have. Risk is a continuous scale, and even mistakes let us learn and adjust. As we learn more, the potential to review and adjust our decisions will increase.
Following up on your decisions is where empathy also plays a role. It should drive the need to measure and follow up on our decision to ensure it is moving us towards our goal. This follow-up is also useful for our model. It feeds our problem stage with more information, and the cycle continues.
How we apply the Empathy Decision Model
Be mindful of how you collect and connect the team with evidence – As UX practitioners, we play an integral part within the bottom two stages of our model: identifying a problem, and determining if the evidence supports it. Therefore, we have an incredible amount of responsibility both in how we carry out our research and how we disseminate insights to the team. UX researchers should use empathy to seek a deeper understanding and clearly identify when we know enough to make good decisions. At Shopify, we also have projects which we call ‘loose threads’. Loose threads are issues, needs, or problems that are repeatedly found through various pieces of research that can appear small at first because they are not necessarily specific to our research questions. They generally spark a feeling of empathy and a concern that they could be indicators of something important that warrants further research. We have the autonomy to channel our empathy and embark on these projects to gather more context for better understanding of the issue. Some of these loose threads have turned into high impact projects at Shopify, or at the very least created implications within existing projects.
Building empathy in teams comes with the duty to manage how that empathy manifests itself in the process of product development. At Shopify, we have found decoupling decision making from the dissemination of findings to be helpful in allowing product teams to go away and reflect on insights. We then review how the findings fit into what we know to mitigate the risk of making rash decisions based on empathy. This distinction means decisions will be made only after considering all the information we have and how much weight to assign to it.
Another responsibility is framing information in order to make sure it isn’t taken out of context after a decision is made. Evidence is based on using information that effectively measures the existence and impact of a problem, and this framing is also important for how this evidence might be referenced later. Commonly, this framing is the way in which the information was gathered. Any discipline that seeks and synthesises information has a responsibility not to let that information be misused or taken out of context, therefore maintaining the frame in which it is relevant. To keep the integrity of the evidence and its context, we encourage our UX practitioners when presented with ‘evidence’ to ask questions like ‘what information is that based on?’, ‘how was that information collected?’, and ‘how do you know that to be true?’.
Where possible, use multiple pieces of evidence to make good decisions – At Shopify, we’ve found that successful decisions are those that incorporate various pieces of evidence elicited from different types of information. As facilitators of information, UX researchers, Data Scientists, and Customer Support Data Analysts work together to answer questions. Not all information holds the same weight, but where possible, we seek multiple sources of information.
This collaborative approach has provided us with the depth of information we need to make good decisions, along with useful decision-making criteria when a clear decision is not present. For example, a quantitative experimental approach may result in no significant differences between the current state and the changed experimental state of a user interface (UI). We review this finding with other information from research and customer support to decide whether making the change is qualitatively a better experience. If so, we will make that decision, because it is the best decision, taking into consideration all the evidence we have. Working together means having the opportunity to rationalise one set of information with another.
Embrace the impact of empathy – We are predisposed to feel different levels of empathy and are conditioned to feel empathy for different things at different degrees, and that’s okay. Work with empathy, not against it. What we should stop doing is expecting empathy to be absolute or objective. It is a bias to think we do not have bias.
As our audience grows, so should our approach to understanding it. One way we have embraced the impact of empathy more successfully is by being intentional about the diversity of our teams—both through hiring and staffing projects. We think about diversity of experience, background, culture, and other dimensions. As the organisation scales, we endeavour to offer more solutions to a growing group of users. It will, therefore, be harder to stay close to all users, contexts, and needs because the effort to gather information about them multiplies. Having a more diverse team doesn’t replace research practices but it increases the chances of including wider viewpoints that should be considered when designing for users.
Another consideration is to put safeguards in place to check in on our biases. In knowing that bias exists, we can influence how we think about information by having tools that sense check our reactions to that information. A method developed by one of our UX researchers was a decision tree that people can use after visiting a business.
This decision tree gives our teams a route to think more deeply about what they observed and stops them going directly from empathy to a decision. It does not try to control or dismiss empathy, rather it works with it to rationalise what has been observed while gathering information.
NEXT STEPS
This is an internal view from our company of the traps that can occur in the decision-making process when empathy is commodified. How our company has developed with empathy at the core of the business is overall beneficial for the work we do, so this is a good problem to have.
If your organisation does not have enough empathy for users, that is the first step. When you have empathy, it must be used at the right level of abstraction in the decision-making process. This is our view, and we expect you, as UX practitioners, to take the model away and gather experiences from inside your own environment. You can then develop, pivot, and reinforce our understanding of empathy in order to use it effectively and responsibly in your organisation.
Figure 6. Shopify business visit decision tree.
Rachel Robertson is a UX Lead at Shopify focused on creating quality product experiences at the intersection of technology and commerce. She has over 10 years of experience in communication design, software product design, and UX management. She is passionate about building inclusive teams, working together to deeply understanding users’ needs and solving complex design problems. rachel.robertson@shopify.com
Penny Allen is a Senior UX Research Lead at Shopify. She has been a Researcher in product for over 10 years working for various companies including the BBC and The Guardian. With a background in psychology, she’s always been motivated to learn deeply about users and work on products that solve real human problems. penny.allen@shopify.com
NOTES
We would like to thank Alëna Iouguina, Ariel Chernin, Austin Dains, Ben Watts, Cassie Kaiser, Cynthia Savard Saucier, Dalia El-Shimy, Dave Newton, David Lennie, Emma Craig, Henry Hammond, Ian Mortier, James Bull, Jen Chow, Michael Patten, and Serena Ngai (Shopify) for their contributions to this paper. With special thanks to Sarah Ebbs, Rebecca Pasiak, and Hilary Dwyer for helping us edit this paper.
The views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of Shopify.
REFERENCES CITED
Arbuckle, N. L. & Cunningham, W. A.
2012 Understanding Everyday Psychopathy: Shared Group Identity Leads to Increased
Concern for Others among Undergraduates Higher in Psychopathy. Social Cognition: Vol. 30, No. 5. pp. 564-583.
Baron-Cohen S., Bowen D. C., Holt R. J., Allison C., Auyeung B. & Lombardo M. V.
2015 The ‘Reading the Mind in the Eyes’ test: complete absence of typical sex difference in ~400 men and women with autism. PLoS One; 10: e0136521.
Bloom, P
2016 Against Empathy: The Case for Rational Compassion. Chapter 3: Doing Good.
Bora E., Gökçen S. & Veznedaroglu B.
2008 Empathic abilities in people with schizophrenia. Psychiatry Res; 160: 23–29.
Bose D., Segui-Gomez, M. & Crandall, J. R.
2011 Vulnerability of Female Drivers Involved in Motor Vehicle Crashes: An Analysis of US Population at Risk. Am J Public Health. 2011 December; 101(12): 2368–2373.
Cameron, C. D.
2017 Compassion collapse: Why we are numb to numbers. In E. M. Seppälä, E.
Simon-Thomas, S. L. Brown, M. C. Worline, C. D. Cameron, & J. R. Doty (Eds.), Oxford library of psychology. The Oxford handbook of compassion science (pp. 261-271). New York, NY, US: Oxford University Press.
Cameron, C. D., & Payne, B. K.
2011 Escaping affect: how motivated emotion regulation creates insensitivity to mass suffering. Journal of Personality and Social Psychology, 100, 1–15.
Decety J & Moriguchi Y.
2007 The empathic brain and its dysfunction in psychiatric populations: implications for intervention across different clinical conditions. Biopsychosoc Med; 1: 22.
Eisenberg N, Eggum ND.
2009 Empathic responding: sympathy and personal distress. In: Decety J, Ickes W, editors. The Social Neuroscience of Empathy. Cambridge: MIT Press; 2009. pp. 71–83.
Esser, P.
2018 https://www.interaction-design.org/literature/article/empathy-how-to-improve-your-designs-by-deveoping-empathy-for-your-target-group
Foucault, M.
2002 The Order of Things: An Archaeology of the Human Sciences. New York: Routledge.
Gutsell, J. N. & Inzlicht, M.
2010 Empathy constrained: Prejudice predicts reduced mental simulation of actions during observation of outgroups. Journal of Experimental Social Psychology Vol. 46, Issue 5, Pages 841-845
Haraway, D. J.
1991 Animal Sociology and a Natural Economy of the Body Politic: A Political Physiology of Dominance. In: Simians, Cyborgs, and Women: the Reinvention of Nature: Routledge, pp. 7–20.
Hodges, S. D. & Myers, M.l W.
2007 Empathy. In R. F. Baumeister and K. D. Vohs (Eds.), Encyclopedia of social psychology (pp. 296-298). Thousand Oaks, CA: Sage.
Hodgson, P.
2017 https://www.userfocus.co.uk/articles/strength-of-evidence.html
Nortona, M., Mochon, D. & Ariely, D.
2012 The IKEA effect: When labor leads to love, Journal of Consumer Psychology: Vol. 22 (3): 453-460.
Slovic, P.
2007 “If I look at the mass I will never act”: Psychic numbing and genocide. Judgment and Decision Making, 2, 79-95.
Tone, E. B. & Tully, E.C.
2014 Empathy as a ‘risky strength’: a multilevel examination of empathy and risk for internalizing disorders. Dev Psychopathol; 26: 1547–1565.
Warrier V., Toro R., Chakrabarti, B., Børglum, A., Grove, J., Hinds, D., Bourgeron, T., & Baron-Cohen, S.
2018 Genome-wide analyses of self-reported empathy: correlations with autism, schizophrenia, and anorexia nervosa. Translational Psychiatry: Vol 8, 35.