Empathy Mapping
A four-quadrant canvas — Says, Thinks, Does, Feels — that captures one person’s experience, where the insight lives in the gaps between the quadrants.
What people say is rarely the whole story. The empathy map exists to expose the distance between what they say, what they do, and what they feel — and that distance is where the design problem actually lives.
What it is
Four quadrants, one person. The insight is in the gaps.
An empathy map is a structured canvas, usually divided into four quadrants — Says, Thinks, Does, and Feels — used to synthesize what a team has learned about a person into a single shared artifact. Where affinity mapping clusters all the research into themes, an empathy map focuses on one person and organizes what you know about them into four dimensions so the team can see them whole.
Its power is not in filling the quadrants; it is in reading across them. The most valuable insight in an empathy map lives in the gaps and contradictions between quadrants: when what someone says differs from what they think, or when what they do contradicts how they feel. A person who says “I know I should save more,” does nothing about it, and feels quiet shame is telling you something no single quadrant could reveal — and it is exactly the kind of tension worth designing for. The map forces a distinction between the said and the thought, the done and the felt, and the distance between them is the gold.
It is a lightweight, fast tool, often used in a workshop as a quick step toward a persona or a point of view. Its simplicity is a strength, but it has one dangerous failure mode: filled from imagination rather than evidence, an empathy map just launders the team’s assumptions into an official-looking artifact.
The four quadrants
Four quadrants, one person. The insight is in the gaps.
Click a quadrant to reveal what goes there, how to fill it honestly, and how to read it in relation to the others. Select SAYS ≠ DOES to see where the most valuable insight lives.
When to deploy it
A fast synthesis tool — not a substitute for research.
Use empathy mapping when
- →You have done initial research and have enough material to characterize a specific person or a well-defined user type based on evidence.
- →You need to quickly align a cross-functional team around a shared, evidence-based understanding of one person before ideation begins.
- →You suspect there is a gap between what your users say and what they actually do or feel, and you want to surface it through structured synthesis.
- →You are on the way to building a persona or a point of view and need a fast intermediate synthesis step that the team does together.
Do not lean on it when
- ×You have no real research. Filling an empathy map from imagination is the single most common abuse of the method — it produces confident fiction dressed as insight.
- ×You need to synthesize a large volume of data across many people. Use affinity mapping, which clusters all the research; an empathy map focuses on one person and is not designed for cross-participant pattern-finding.
- ×You treat the map as a final deliverable. Its value is the shared understanding it builds and the tensions it surfaces, not the artifact. Do not mistake a filled canvas for a finished insight.
The honest limit: an empathy map is only as true as the research beneath it, and its most valuable output — the Feels quadrant and the Says-vs-Does tension — is inferred, not observed. Done carelessly or from assumption, it reassures the team with a tidy picture instead of challenging them with a real one.
How it works
Five moves, from raw research to named tension.
Draw the canvas.
Four quadrants, Says, Thinks, Does, Feels, around a central representation of the person. Some versions add Pains and Gains at the bottom; the four quadrants are the core. The canvas itself is trivial — a whiteboard, a digital template, a piece of paper with lines drawn on it. What matters is the evidence and the honest reading.
Populate each quadrant from evidence.
Fill each quadrant with entries drawn from actual research: quotes from interviews for Says, observed behavior for Does, and careful inference for Thinks and Feels. Every entry should trace to something real. The discipline of pointing to the source for each note is the main guard against filling the map from imagination.
Spend the most time on Feels.
The natural tendency is to fill Says and Does quickly (they are observable and stated) and stop. Budget explicitly for Feels. Ask: what is the emotional register of this person? What did they not say, but that was present in the tone? The emotion they named is often not the deepest emotion. Read for what they avoided.
Read across the quadrants for contradictions.
The valuable work is comparative, not descriptive. Where does Says diverge from Does? Where does Feels explain a behavior that Says would not predict? Hunt the contradictions. A map with no tension is usually a map filled from assumption — real research rarely produces a person with no internal conflict.
Name the tension worth designing for.
Identify the central gap — usually the contradiction between what the person says or does and what they feel — and treat it as the thing the design must address. The empathy map is a thinking tool on the way to a framed design problem, not a destination. It feeds a point of view, a How Might We question, or a persona — what comes next is where the synthesis is used.
Best practices
What good looks like — and what prevents it.
When it goes well
- ✓Every entry traces to real research evidence — a quote, an observed behavior, or a carefully inferred emotion drawn from the emotional register of the interview.
- ✓The map surfaces a genuine tension — usually a contradiction between what the person says and what they do, explained by what they feel.
- ✓The Feels quadrant carries real emotional weight: not a single bland word but a specific, inferred emotional truth that the person may never have named.
- ✓The team reads across the quadrants for gaps, rather than just filling each one in isolation and declaring the exercise complete.
- ✓It is used as a stepping stone to a point of view, a How Might We question, or a persona — not filed away as a finished deliverable.
The mistakes, and how to avoid them
Filling it from assumption.
The single most common abuse. An empathy map built from imagination dresses the team's guesses up as insight. Ground every quadrant in evidence, and keep the source research visible during the session so each entry can be pointed to.
Neglecting Feels, or filling it lazily.
A Feels quadrant with one word ("frustrated") has not done the work. The emotional truth the method exists to hold is specific, inferred, and usually un-stated. Budget the session's discussion time for Feels, and read carefully for what the person did not name.
Treating all four quadrants as equal.
They are not. The gap between Says and Does is the gold. A map that fills all four neatly but surfaces no tension has missed the point. The contradictions are the output, not the completeness.
Producing a reassuring map.
A tidy empathy map that confirms what the team already believed is a warning sign. A good one should contain something uncomfortable. If the map doesn't surprise anyone, it was probably filled from assumption rather than evidence.
Logistics
Running the session from evidence to shared understanding.
Empathy mapping is quick and collaborative by design. It works well as a short, focused session where the team fills the quadrants together from the research, then steps back to read across them. Its speed is a feature: the session should take 45–90 minutes for a single well-researched person. Do not over-engineer it.
Work from real research, visibly
Keep the source research — interview quotes, observation notes — in front of the team as they fill the map, so each entry can be traced to evidence. A sticky note that cannot be sourced to the research is an assumption. The discipline of pointing to the source is the main guard against the imagination-filled map.
One map, one person
An empathy map is for a single person or a single well-defined user type. Trying to map 'our users' in general produces a blurred average that is true of nobody in particular. If you have distinct types who behave very differently, make a map for each.
Spend the most time on Feels and the gaps
Says and Does fill quickly because they are stated and observable. The session's most valuable time is in the Feels quadrant and in the conversation about contradictions across quadrants. Budget explicitly: reserve at least a third of the session for Feels and the cross-quadrant reading.
Build shared understanding, not just the artifact
The map is not the output. The shared understanding the team builds by filling it together is the output. Every team member who fills a sticky note is building their own model of the person. The conversation about which notes belong in Feels — and which ones are assumptions — is where the value is generated.
Remote: works with light facilitation
Empathy mapping works well remotely on shared digital canvases (Miro, Mural, FigJam, named as common examples). The key facilitation requirement is enforcing the silent individual fill phase before discussion — without it, the dominant voice in the room defines the map before the evidence is examined.
AI and this method
AI fills Says and Does well. Thinks and Feels are inferences — and the inference is where the method lives.
Toggle between modes to see which quadrants AI handles reliably — and where it produces a plausible, emotionally flat map that misses the tension worth designing for.
In-depth example
The same research. Two maps. One insight that only one of them could make.
A financial-services team is designing a retirement-planning tool for people in their early thirties. Research is complete. Both versions map the same person from the same interview transcripts; only the method of synthesis differs. Toggle between them to see where the reframing insight comes from.
Scenario
Designing a retirement-planning tool for people in their early thirties
A financial-services team has run depth interviews with their target user: a 32-year-old professional who uses the product intermittently and has never seriously engaged with retirement planning. The team builds an empathy map from the interview evidence. The goal is a shared understanding of this person before ideation begins.
SAYS
OBSERVED- —"I know I should be saving more for retirement."
- —"I'm pretty responsible with my money day-to-day."
- —"I'll properly sort out my pension contributions next year."
Direct quotes from the interview. Stated with confidence and no apparent discomfort.
THINKS
INFERRED- —"Retirement is so far away I can't really picture it."
- —"I don't really understand how the pension numbers work."
- —"Other things — rent, travel, saving for a house — feel more real."
Inferred from the texture of the conversation — what the person circled around without stating directly.
DOES
OBSERVED- —Avoids opening retirement account statements when they arrive.
- —Has not increased contributions in three years despite two pay rises.
- —Opens the planning tool occasionally, looks at the dashboard, closes it without acting.
Observed behavior from behavioral data and self-reported behavior confirmed by interview detail.
FEELS
INFERRED- —Quiet shame whenever retirement comes up — a feeling of already being behind.
- —Genuine anxiety when confronted with pension numbers she doesn't understand.
- —A background sense of failure that she rarely names, and doesn't mention in the interview.
Carefully inferred from tone, hesitation, what was avoided, and the emotional register of the conversation.
The Says-vs-Does gap — and what it revealed
SAYS: “I know I should save more.” DOES: avoids everything to do with it. FEELS: quiet shame that makes every interaction feel like a reminder of failure.
The contradiction between SAYS and DOES was stark — but it was FEELS that explained it. The user was not failing to save because she lacked information. She already knew what she should do. She was failing to save because every interaction with retirement planning made her feel like she was confronting evidence of her own inadequacy — and she avoided it the way one avoids anything painful. The insight reframed the product entirely: the problem was not a lack of information; it was a shame loop that made engagement feel worse than avoidance. The design pivoted from providing more data to making progress feel achievable and non-threatening. The opening screen changed from a projected retirement balance (the number that felt like an accusation) to a small, completable step that felt like winning rather than losing.
What made this possible
The insight came from reading across the quadrants, not from any single one. SAYS and DOES in isolation looked like a simple consistency problem. FEELS, carefully filled from the emotional register of the interview rather than from the words the person used, explained the contradiction. The team spent twice as long on FEELS as on the other three quadrants combined — and that time produced the reframing insight the design pivoted on.
Frameworks
Where empathy mapping shows up.
Empathy mapping is a fast synthesis tool, so it maps to the moments in each framework where the task is building shared understanding of a specific person before design decisions are made.
Related methods
What to combine with empathy mapping.
Sources & further reading
The work behind this method.
Gamestorming
Dave Gray, Sunni Brown, and James Macanufo (2010)
The book that popularized the empathy map format and remains the best source on the structure and intent of the four-quadrant canvas. Gray's framing of the method emphasizes the Feels quadrant and the reading-across discipline — making the gaps between quadrants the productive output, not the completeness of each.
This Is Service Design Thinking
Marc Stickdorn and Jakob Schneider (2011)
Situates empathy mapping within service design practice, where understanding the emotional experience of a user at each service touchpoint is central to the design task. The emphasis on the emotional and contextual dimensions of the person — what they feel throughout a service journey — is directly relevant to the Feels quadrant.
Interviewing Users
Steve Portigal (2013)
The definitive guide to gathering the research — and the emotional signal — that a good empathy map depends on. Portigal's techniques for reading tone, silence, and what is not said are exactly what separates a Feels quadrant filled with real emotional truth from one filled with surface words. The quality of the interview determines the quality of the map.