Empathy Mapping
✓ four quadrants, every entry traceable to evidence
On this page
- TL;DR
- What is an empathy map?
- Why empathy maps go wrong
- The four empathy map quadrants, properly
- How to build an empathy map from research
- Keep the empathy map connected to current research
- Empathy map vs. persona vs. journey map
- When to use an empathy map
- How to run an empathy mapping session
- Common empathy mapping mistakes
- Build the map from evidence, not imagination
TL;DR
An empathy map organizes what you know about a user into four areas: what they say, what they think, what they do, and how they feel.
The format is simple. The problem is how easily it turns assumptions into something that looks like evidence. A team gets in a room, fills four boxes from memory, and leaves with a polished artifact that may say more about the team than the customer.
A useful empathy map starts somewhere else: with real research. Use actual quotes for what people say, observed behavior for what they do, clearly labeled inferences for what they think and feel, and a source attached to every important entry.
Build one in four steps: pick one person and one situation, gather the evidence, sort it into the quadrants, then look for the gaps and contradictions. Those last two are often where the real insight is.
You can also start with our empathy map template and fill it with evidence from your own research rather than workshop guesses.
What is an empathy map?
An empathy map is a visual framework for organizing what a specific user experiences in a specific situation.
The classic version has four quadrants arranged around the user.
- Says. What the person actually says out loud, ideally in their own words. “I just export it and do it in a spreadsheet.”
- Thinks. What may be going through their head even if they do not say it directly.
- Does. Observable behavior: what they click, avoid, repeat, work around, or abandon.
- Feels. Their emotional state during the experience, especially when that emotion changes.
Pains and gains
Many empathy maps also include Pains and Gains around the outside. Pains capture frustrations, obstacles, and risks. Gains capture what the person wants, values, or defines as success.
The value of the format is not that four boxes magically create empathy. It is that they force you to separate things teams often blur together.
What someone says and what they do can be very different. Someone might tell you reporting works fine, then spend 20 minutes rebuilding the same report manually every Friday. That contradiction is more useful than either observation on its own.
Why empathy maps go wrong
Empathy mapping often happens in a workshop. Six people get in a room. Someone writes “wants things to be easy” under Thinks. Someone adds “frustrated by complexity” under Feels. Everyone agrees. The map looks plausible.
But none of it necessarily came from a user.
The problem is not the workshop itself. Collaborative synthesis can be valuable. The problem is treating ideas generated in the room as if they came from research. Three things tend to go wrong.
Assumptions start looking like evidence
An invented quote in the Says box looks exactly like a real quote once it is on a sticky note.
There is a meaningful difference between “I can't tell if it saved” from a participant during a usability session, and the same sentence written by a designer imagining what a participant might say.
If the map does not preserve that distinction, assumptions can quietly turn into customer insight.
The map becomes generic
When teams work from memory, the same themes appear again and again. Users want things to be:
- Easy
- Fast
- Clear
- Reliable
- Intuitive
All of those may be true. None tells you much about what to build or fix. Specific research produces specific empathy maps.
Several people get flattened into one fictional person
Combining every customer you have ever spoken to into one map can create a user who does not really exist.
One participant is a new administrator. Another is an experienced power user. Another is trying to cancel. Their emotions, behaviors, and expectations may all be reasonable, but they belong to different situations.
The tighter the situation, the more useful the map becomes.
The four empathy map quadrants, properly
The boxes look equal on the template, but methodologically, they are not.
Says
Use real participant language. If an entry appears in Says, you should be able to trace it back to a transcript, recording, ticket, or another source where a real person actually said it.
This is where user interviews, support conversations, surveys, and other customer feedback become especially useful.
Avoid rewriting everything into cleaner product language. The participant's phrasing is part of the evidence.
Thinks
Thinks is usually inference. You rarely have direct access to what another person is thinking, so treat this box differently from Says.
Instead of presenting interpretations as facts, label them clearly. Specific inferences can be examined and tested. Generic ones cannot.
Does
Does should be grounded in observed behavior whenever possible. Useful sources include:
- Usability testing
- Product analytics
- Session recordings
- Contextual research
- Support behavior
- Direct observation
This quadrant often gets less attention because it requires actually seeing what someone did rather than asking what they remember doing.
It is also where some of the best contradictions appear. A user can say a process is easy while repeatedly backtracking through it. Both observations belong on the map.
Feels
Avoid reducing the emotional layer to happy, confused, or frustrated. Instead, look for the moment the emotion changes.
Did someone begin confident and become uncertain after a warning message? Did irritation turn into resignation after the third failed attempt? Did anxiety disappear once they saw a confirmation screen?
Those transitions are much more actionable than assigning one emotion to an entire experience.
Pains and gains
If you add Pains and Gains, derive them from the evidence already on the map. A pain should connect to something the participant said, did, or experienced. A gain should connect to an outcome they actually value.
That keeps these sections from becoming another place for the team to brainstorm what customers probably want.
How to build an empathy map from research
Step 1: Pick one person and one situation
Do not map “our users.” Choose one meaningful audience in one meaningful moment, like:
- A new administrator in their first week trying to get the rest of the team set up.
- An experienced customer trying to rebuild a report after a product update.
Emotions and behavior are situational. Someone can feel completely confident in one workflow and lost in another.
If you already have research-backed personas, they can help define who you are mapping. Just remember that the empathy map inherits the quality of the persona underneath it. A persona built largely from assumptions does not become evidence simply because you put it through another framework.
If you are still figuring out who you need to understand, the broader customer research guide can help you define the audience before you start mapping.
Step 2: Pull the evidence
Gather research from people who actually experienced the situation you are mapping. Useful sources include:
- Interview transcripts. Especially useful for Says and for context that helps you interpret Thinks.
- Usability sessions. Strong for Does and Feels because you can see hesitation, retries, backtracking, and reactions as they happen.
- Support tickets. Useful for pain points expressed in the customer's own language, often close to the moment of frustration.
- Sales and churn conversations. Helpful for understanding expectations, unmet needs, and why someone moved toward or away from the product.
- Product analytics. Useful for behavior at scale: where people actually go, stop, return, or drop out.
The number of participants you need depends on the situation and how varied the audience is. A focused set of well-matched participants can be enough to reveal meaningful patterns, but avoid treating one person's experience as representative of an entire segment.
If new sessions are still producing substantially different experiences, you probably need more evidence before you turn the map into a shared reference point.
Step 3: Sort the evidence and keep the source attached
Every important entry should include a way back to the original evidence. That might be:
- A participant or session ID
- A timestamp
- A ticket number
- A survey response
- A link to the transcript or recording
This is what makes the map defensible. It also makes it reusable. If someone challenges an entry six months later, you can check the evidence instead of arguing about who remembers the workshop correctly.
A research repository makes this much easier because the map can point back to source material rather than becoming an isolated whiteboard.
Mark interpretation separately from direct evidence too. Someone reading the map should be able to tell which entries are quotes or observed actions and which are your team's inference.
Step 4: Look for gaps and contradictions
A full empathy map is not the goal. Insight is.
Look for places where Says and Does conflict. That contradiction may tell you more than either line alone.
Look for empty quadrants too. An empty Feels box does not necessarily mean emotion is irrelevant. It may mean your existing research never captured it. That is a useful gap.
Also look for things you expected to see but did not. If your team just launched a major feature and nobody in the relevant situation mentions or uses it, that absence is information worth investigating.
The map should leave you with both findings and questions. Our research synthesis guide goes deeper on turning patterns like these into evidence-backed findings your team can act on, and affinity mapping is a useful next step when you have more raw evidence than structure.
Keep the empathy map connected to current research
An empathy map describes a moment in your customer experience. That means it can go stale. Products change. Workflows change. Customers develop new habits. Problems get fixed and new ones appear.
Rather than treating the map as a finished artifact, connect it to the research underneath it. When the source interviews, usability sessions, and customer feedback live in a searchable repository, it becomes much easier to revisit the map as new research arrives.
Teams practicing continuous discovery have an advantage here because customer evidence is already being collected on a regular basis.
You do not necessarily need to rebuild the whole map every quarter. Revisit the evidence, check whether the major patterns still hold, and update the parts that have changed. The UX research repository guide covers how to build the broader system around that research.
Empathy map vs. persona vs. journey map
Empathy maps, personas, and journey maps often appear in the same workshops, but they answer different questions.
An empathy map focuses on what one user experiences in one situation. It is especially useful for zooming into thoughts, behaviors, emotions, and contradictions.
A persona represents a recurring type of user across situations. It typically captures broader goals, needs, behaviors, and context.
A journey map follows an experience over time. It shows the stages someone moves through, what happens at each stage, and where the process breaks down.
They can work together. A journey map might reveal that onboarding week two is the biggest source of frustration. An empathy map can then zoom into that stage to understand what the customer says, thinks, does, and feels while the problem is happening.
When to use an empathy map
An empathy map can be useful when you need to turn qualitative research into a shared picture the team can quickly understand. Good use cases include:
- Bringing a cross-functional team closer to customer evidence
- Preparing for a design sprint or workshop
- Zooming into a difficult moment discovered through research
- Comparing how different segments experience the same situation
- Onboarding a new team member to a customer problem
- Organizing evidence before deeper synthesis
When it is a poor fit
It is a poor fit when you need:
- A substitute for talking to customers
- Quantitative evidence
- A decision between two interface designs
- A view of an entire journey over time
- One map that supposedly represents several very different user types
If the question is which design works better, run usability or preference research. If the question is how common a problem is, use a method designed to measure prevalence.
The UX research methods guide can help you choose the right method instead of asking an empathy map to answer a question it was never designed for.
How to run an empathy mapping session
Building the map collaboratively can be useful because the discussion itself exposes assumptions and gives more people direct contact with the research. The sequence matters.
Give people the evidence first
Do not ask a group to walk into the session cold and start filling boxes. Share the relevant transcripts, clips, notes, or findings in advance.
If the team has not engaged with the research, the workshop will naturally fall back on assumptions.
Start with the participant, not the template
Open with a short clip or quote from the research.
Even two minutes of a real participant struggling with a workflow can reset the room from “What do we think users feel?” to “What did we actually see?” That distinction changes the rest of the conversation.
Fill Says and Does before Thinks and Feels
Start with what you can source directly. Add quotes to Says. Add observed behaviors to Does. Only then move into Thinks and Feels.
That order reduces the temptation to invent an emotional story first and then search the evidence for something that supports it.
Mark assumptions as assumptions
If someone adds an idea the research does not support yet, do not necessarily throw it away. Label it.
An unsupported hypothesis can still be useful if everyone knows it is a hypothesis. It becomes dangerous when it is visually indistinguishable from customer evidence.
End with what you do not know
Before wrapping up, ask:
- Which quadrants have weak evidence?
- Which entries are inference rather than observation?
- Where do participants contradict one another?
- What surprised us?
- What do we need to investigate next?
That list may be more valuable than the finished map because it gives you a ready-made research plan. If you need to design the next interview study, our guide to writing a user interview discussion guide can help turn those gaps into useful questions.
Common empathy mapping mistakes
Filling every box
Empty space is not failure. An empty quadrant can tell you exactly where your research is thin. Filling it with assumptions makes the map look more complete while making it less trustworthy.
Putting invented quotes in Says
If nobody said it, it does not belong in Says. Move it to Thinks as an inference or label it as a hypothesis.
Mapping “the user”
Choose one audience and one situation. If several user types experience the same workflow very differently, build separate maps.
Treating the map as a deliverable
An empathy map is primarily a thinking and synthesis tool. Its value comes from what the team learns while building and discussing it, not how polished the final diagram looks. Do the analytical work before worrying about making it presentation-ready.
Never updating it
A map built from research two product versions ago may no longer describe the current experience. Connect the map to living evidence and revisit it when the underlying workflow changes.
Using it to settle a design debate
An empathy map can help explain why a design problem matters. It cannot tell you which of two designs performs better. Test the designs with people.
Build the map from evidence, not imagination
Empathy maps are useful because they compress a lot of qualitative evidence into something a team can understand quickly. That is also what makes them risky. Four neatly labeled boxes can make assumptions feel more certain than they are.
The fix is simple: keep the evidence attached. Use real quotes. Preserve observed behavior. Mark inferences as inferences. Leave gaps visible. Link every important entry back to the session, ticket, or response it came from.
Great Question brings customer research, participant recruitment, studies, and a searchable research repository into one place, so the evidence behind an empathy map does not have to disappear into a static workshop board.
Start with the empathy map template, then fill it with what your customers actually said and did.
Frequently asked questions
What is an empathy map?
An empathy map is a visual framework that organizes what a user says, thinks, does, and feels in a specific situation. Many versions also include pains and gains. The goal is to help a team build a shared understanding of the user's experience based on research evidence.
What are the four quadrants of an empathy map?
The four classic quadrants are Says, Thinks, Does, and Feels. Says should contain direct participant language, while Does captures observable behavior. Thinks and Feels often involve interpretation and should be clearly distinguished from direct evidence.
How is an empathy map different from a persona?
A persona describes a recurring type of user across situations, including broader goals, behaviors, and context. An empathy map focuses on what a specific user or segment experiences during one particular situation.
How many users do you need for an empathy map?
There is no universal number. The right amount of research depends on how narrowly the audience and situation are defined and how much variation exists between participants. A small, well-matched qualitative sample can reveal meaningful patterns, but avoid treating the experience of one or two people as representative of an entire segment.
Can you make an empathy map without research?
You can use the format to capture assumptions, but label the result clearly as an assumption map or hypothesis. Do not present brainstormed Says, Thinks, Does, or Feels entries as customer evidence until they have been validated with real research.
