An empathy map is a simple visual tool that helps you organize what you know about a user into four categories: what they say, think, do, and feel. It sounds basic, but a well-built empathy map can reveal contradictions, unmet needs, and design opportunities that raw interview notes never would.

Why Empathy Maps Work

After conducting user interviews, most teams face the same challenge: they have pages of notes, hours of recordings, and a vague sense of what they learned, but no clear way to turn it into action. Empathy maps solve this by forcing structured synthesis.

The four-quadrant format works because it separates different types of evidence. What a person says they do and what they actually do are often very different things. What they feel and what they think are related but distinct. By placing observations into specific quadrants, patterns and contradictions become visible that would stay hidden in chronological interview notes.

Dave Gray, who popularized the empathy map at XPLANE, designed it specifically for this purpose: to help teams move from "we talked to some users" to "we understand these users well enough to design for them."

The four-quadrant empathy map with pain and gain zones

The Research and Evidence

Nielsen Norman Group recommends empathy maps as a synthesis tool that helps teams move from raw interview data to actionable design insights, particularly when teams need shared understanding of user needs before ideation. The value is not in the map itself but in the act of building it: the structured conversation forces team members to distinguish between what they observed and what they inferred, and to surface disagreements about user motivations that would otherwise stay hidden.

In organizational contexts, empathy mapping has proven valuable beyond product design. DTGroup published a case study of a pharmaceutical company undergoing restructuring that used empathy mapping during the integration process. The exercise surfaced misalignments between departments that leadership had not identified through conventional methods, leading to a revised integration plan that addressed actual employee concerns rather than assumed ones.

The Four Quadrants

Says

Direct quotes from interviews, support tickets, reviews, or social media. Use the person's exact words whenever possible. Paraphrasing loses nuance. If a user said "I literally dread opening that app every Monday morning," capture it exactly like that. The word "dread" tells you something that "doesn't enjoy using the app" does not.

Good examples:

"I spend half my morning just figuring out what I'm supposed to work on."
"I've given up trying to customize it. I just use the defaults."
"My manager asks me for this report every week and I have to build it from scratch every time."

Thinks

What is going on inside the user's head? This requires inference from their behavior, tone, and context. You cannot directly observe thoughts, but you can make reasonable inferences when a user hesitates before answering, qualifies a statement, or contradicts something they said earlier.

Good examples:

"Wonders if there's a faster way but doesn't have time to look for one."
"Suspects the tool can do more but feels intimidated by the advanced features."
"Worries about making a mistake that will be visible to the whole team."

Does

Observable actions and behaviors. What do they actually do when they encounter the problem? This is the most objective quadrant because it is based on what you can see rather than what someone tells you. Contextual observation is the best source for this quadrant.

Good examples:

"Opens three different spreadsheets to find the information she needs for one task."
"Screenshots the dashboard and pastes it into Slack rather than sharing the link."
"Keeps a Post-it note on the monitor with the steps for the monthly report."

Feels

Emotional states and reactions. Look for emotions in facial expressions, body language, tone of voice, word choice, and the intensity behind statements. People rarely say "I feel frustrated." They sigh, they laugh nervously, they say "it's fine" in a tone that clearly communicates it is not fine.

Good examples:

"Anxious about making errors because there is no undo button."
"Relieved when the task is finally done, but exhausted by the process."
"Proud of the workaround she built, but resentful that she had to build it."

The Pains and Gains Extension

Many teams add two sections below the four quadrants:

Pains: Frustrations, obstacles, fears, and risks the user faces. What keeps them up at night? What do they want to avoid? What are the consequences of failure?
Gains: Goals, desires, and measures of success. What does a good outcome look like? What would make their day? What would they brag about to a colleague?

Pains and Gains help bridge empathy mapping into problem definition. They connect user emotions to concrete design opportunities that feed directly into the Define stage.

Step-by-Step: Building an Empathy Map

Step 1: Gather Your Raw Material

Collect everything you have from your empathy research: interview transcripts, observation notes, survey responses, support ticket logs, app reviews, forum posts. Even 3 to 5 thorough interviews provide enough material for a useful empathy map.

Step 2: Choose Your Scope

Decide whether you are mapping a single user or a user segment. Individual maps preserve nuance and are best when you have done deep interviews with specific people. Segment maps aggregate multiple users and are better for team alignment and persona development.

If you are creating segment maps, start by building individual maps first, then combine them. This prevents you from averaging out the interesting edge cases.

Step 3: Fill Each Quadrant

Go through your research chronologically. For each observation, quote, or insight, place it in the appropriate quadrant. One insight per note. Be as specific as possible.

A common mistake is filling quadrants too abstractly. "Gets frustrated" belongs on a mood board, not an empathy map. "Gets frustrated because the export takes 3 minutes and there is no progress indicator, so she does not know if it is working or frozen" is an insight you can design for.

Step 4: Look for Contradictions

The most valuable part of empathy mapping is finding where the quadrants contradict each other. These contradictions reveal the deepest insights:

Says vs Does: "I always read the documentation" but the Does quadrant shows they google the answer instead. This tells you the documentation is not working, regardless of what users claim.
Thinks vs Feels: "Thinks the tool is powerful and capable" but "Feels intimidated and avoids advanced features." This tells you the problem is not capability but approachability.
Says vs Feels: "Says everything is fine" but "Feels resigned and has stopped expecting improvement." This tells you satisfaction surveys are misleading for this user group.

These contradictions are gold. They point to problems users cannot or will not articulate directly, which means your competitors are probably missing them too.

Step 5: Extract Needs

Translate your patterns and contradictions into user needs. Frame them as verbs:

"Needs to feel confident that her work is saved before closing the app."
"Needs to understand what happened when something goes wrong, without technical jargon."
"Needs to accomplish the weekly report in under 15 minutes so it does not eat into her actual work."

These needs become the raw material for How Might We questions in the Define stage.

Individual vs Aggregate Empathy Maps

Individual empathy maps capture one specific person's perspective. They preserve the richness and specificity of a single interview. Use them when you want to maintain individual nuance and when you will reference specific users throughout the project.

Aggregate empathy maps combine observations from multiple users into a single map representing a user segment. They are more useful for team alignment, persona creation, and stakeholder communication. The risk is that aggregation smooths out the extreme cases that often contain the most interesting design opportunities.

Best practice: build individual maps first, then create aggregate maps while flagging important outliers. If one user out of eight describes a completely different experience, that outlier might represent an underserved segment worth investigating.

Common Mistakes

Filling Thinks and Feels without evidence. These quadrants require inference from observed behavior, not imagination. If you have no evidence for what a user thinks or feels about a topic, leave it blank. Blank spaces are signals that you need more research, not embarrassments to fill with guesses.
Making it too abstract. Every entry should be specific enough that a designer could act on it. Compare "is frustrated" with "is frustrated because the search returns 200 results with no way to filter, so she scrolls manually every time."
Doing it once and filing it away. Empathy maps are living documents. Update them as you learn more through prototyping and testing. The map you create after testing will be significantly richer than the one you created after initial research.
Projecting your own experience. The most dangerous entries in an empathy map are the ones that come from the team's assumptions rather than from user evidence. If you catch yourself writing "probably thinks..." without a specific observation to back it up, stop.
Skipping the synthesis step. The map itself is not the deliverable. The needs and insights you extract from it are. Treat the map as a tool, not an artifact to hang on the wall.

Empathy Maps in Practice

A product team at a mid-size SaaS company used empathy mapping after interviewing 12 customers who had churned. The Says quadrant was full of polite exit-interview language: "We just went in a different direction." But the Does quadrant told a different story: 9 of 12 had stopped logging in weeks before they officially cancelled, and 7 had exported their data more than a month before cancellation.

The contradiction between Says ("decided to switch recently") and Does ("started disengaging months ago") revealed that churn was not a sudden decision but a slow fade. This insight changed the team's retention strategy from exit offers to early-warning engagement campaigns, a solution that never would have emerged from the surface-level exit interview data alone.

AI-Assisted Empathy Mapping

AI tools can accelerate empathy mapping by analyzing interview transcripts and suggesting entries for each quadrant. Design Thinker Labs generates empathy maps from your project context, providing a structured starting point that you refine with your own observations and direct user interactions. This is especially useful for teams that are new to empathy mapping and benefit from seeing a worked example before building their own.