Empathy map · User research · UX
An empathy map example, from interview notes to your next questions
By T. Krause
An empathy map is only as good as the research behind it. Here is one made-up interview with a first-time home buyer, turned line by line into Says, Thinks, Does and Feels, and the three contradictions it shows for the next interview.
Here is a short empathy map example. A first-time home buyer says "Can we afford it?", thinks rent is wasted money, checks listings every night and feels overwhelmed. To create one of your own, take your notes from research on one user or user type, sort them under Says, Thinks, Does and Feels, then look for the places where the boxes disagree. Those disagreements become the questions for your next interview.
This guide follows that buyer from a made-up interview to a finished map. Sorting is where most of the judgement goes, so it gets a table of its own, line by line.
We make Prompt Board, and our empathy map template opens on this same home buyer, with the four boxes ready to overwrite. Sticky notes on a wall work too.
The finished map
Everything in this example is invented: the buyer, the interview and the team. The map comes from a made-up team building an app for first-time buyers. It's the same eight notes you'll find on the sample board.
| Box | Notes |
|---|---|
| Says | "I don't get the fees". "Can we afford it?" |
| Thinks | Rent is wasted money. Rates might drop. |
| Does | Checks listings nightly. Asks parents for advice. |
| Feels | Overwhelmed. Excited and anxious. |
Nielsen Norman Group's guide to empathy mapping calls an empathy map "a collaborative visualization that captures what we know about a particular user or about a user segment". It also says the map is deliberately not chronological: "Whereas a journey map follows a sequence of steps, an empathy map captures a mindset."
Read those eight notes together and a mindset does appear. This person wants out of renting, can't tell what buying would cost, and looks every night anyway.
How the interview became those notes
The team's researcher talked to Nadia, 31, who rents a flat with her partner. Here is a stretch of the transcript.
"I don't get the fees. The agent sent a list and I couldn't tell which ones we'd actually pay. Can we afford it? That's what I keep coming back to. I look at listings every night after dinner, it's a bit of a habit now. And every month the rent goes out and I think, well, that money's gone. But my dad says rates might drop, so maybe we should wait. Honestly, I go from excited to anxious in about five minutes."
During that answer, she also picked up the agent's fee list, looked at it for a moment and put it face down on the table.
The team turn that into notes like this.
| What the transcript has | Note | Box | Why that box |
|---|---|---|---|
| "I don't get the fees." | "I don't get the fees" | Says | Her own words, kept in quotation marks |
| "Can we afford it?" | "Can we afford it?" | Says | Her own words again |
| "I look at listings every night after dinner" | Checks listings nightly | Does | A behaviour, even though she told it rather than showed it |
| "my dad says rates might drop" | Asks parents for advice | Does | Talking to her father is something she did |
| "my dad says rates might drop, so maybe we should wait" | Rates might drop | Thinks | She has taken on his view and is weighing it |
| "that money's gone" | Rent is wasted money | Thinks | The researcher's wording for a belief she implied |
| "from excited to anxious" | Excited and anxious | Feels | She named the feelings herself |
| Puts the fee list face down | Overwhelmed | Feels | The researcher's reading of something she did, and the least certain note on the map |
One sentence can produce two notes, as her father's advice does here. And the last row is a judgement call. Nielsen Norman Group's rule is that every note "should trace back to a specific research artifact", and that a note you can't source "is an assumption, and it should be labeled as one". "Overwhelmed" has a source, the moment with the fee list, but the word is the researcher's. So the team would check it in the next interview.
Where the awkward notes go
These are the boundaries we'd expect a team to argue about at the board.
Says or Thinks. Use Says for words you can quote. Use Thinks for what goes through the person's mind, whether they said it or you inferred it. A tidied-up paraphrase of a quote belongs in neither until you've decided whether it's a quote or your reading of one.
Says or Does. "I look at listings every night" is a report of behaviour. It goes in Does, and it's weaker evidence than watching someone do it. What people report and what they do can differ, so use recordings or analytics where you have them.
Thinks or Feels. A sentence with "that" after it is usually a thought: she feels that rent is a waste. A sentence with an emotion word in it is a feeling. Nielsen Norman Group suggests writing Feels as "an adjective plus a short phrase that explains it". The sample board keeps its notes short, but "Overwhelmed: can't tell which fees she'd pay" tells a team more than one word does.
A note that fits two boxes. Don't spend long on it. The same guide's advice is that "if an item could fit into multiple quadrants, pick one and move on".
Read the contradictions
A map where every box agrees hasn't told you much yet. Nielsen Norman Group says contradictions between quadrants "are worth investigating, because they often point to something that your users did not articulate during research". Nadia's map has three, and each one gives the team a question.
| What disagrees | What it might mean | Next interview question |
|---|---|---|
| Says "Can we afford it?", Does: checks listings nightly | She's shopping with no top price in mind | "Have you worked out the most you could pay? What stopped you?" |
| Thinks rates might drop, Does: checks listings nightly | She talks about waiting and still looks every night | "What would have to happen for you to stop waiting?" |
| Feels overwhelmed by fees, Does: asks her parents | She may trust family over the agent on money | "Whose answer about the fees would you believe, and why?" |
Notice that none of these is a feature idea yet. A fee calculator might be the right answer, or the problem might be that nobody she trusts has explained the fees to her. The team can't tell from one interview, and the map shows them what to ask to find out.
Gaps count too. If you can't fill a box, the same guide calls that "a strong signal that you need more research before moving further in the design process".
Make one from your own research
- Name one person or one user type at the top. Nielsen Norman Group says to "start with a one-to-one mapping", one user or persona per map. Make a second map for a second type.
- Bring the research into the room. That means transcripts, recordings, support tickets or field notes. Read them before you write anything.
- Write notes alone first. Nielsen Norman Group warns that "people should not see each other's notes in this phase", because early opinions anchor the group. On a shared board, give people a few minutes before anyone starts talking.
- Add a source to each note. A short tag such as "Interview 2, 14:30" is enough. Mark anything without a source as an assumption.
- Group, then look across. Merge the duplicates inside each box and name the clusters. Then read across the boxes for contradictions, and write each one as a question.
- Date it and keep going. Put the date and the number of interviews on the board. A map from one interview is a set of leads, and it should change after the next one.
One person or a whole segment
An empathy map can be about one real person or about a group. Nielsen Norman Group's terms are one-user maps and aggregated maps, which "represent a user segment rather than one particular user". They are usually made by combining individual maps from users who behave alike.
We'd make one-user maps first, even if you want a segment in the end. They keep the evidence visible. It's easy to write "Buyers are overwhelmed" on a segment map when only one buyer turned a fee list face down. If five individual maps show the same thing, you've earned the summary.
Empathy map, persona or journey map?
These three get used together and get mixed up, and each answers a different question.
| Format | The question it answers | Good moment to make it |
|---|---|---|
| Empathy map | What's going on in this person's head, and where does it not add up? | Right after a round of interviews |
| User persona | Who are we building for, and what do they want? | When the team needs one shared profile to argue with |
| User journey map | What happens step by step, and where does it go wrong? | When you're fixing a process with a clear start and end |
They can also work in that order. Several empathy maps can feed a persona, and a persona gives a journey map its main character.
Where the empathy map came from
The oldest write-up we found is an undated post on the Gamestorming site, with reader comments from November 2009. It calls the map "one of XPLANE's methods for understanding audiences", shows a picture uploaded by dgray_xplane, and notes that Alex Osterwalder featured it in Business Model Generation. We didn't find a dated source that goes further back, so we can't give you a year of invention.
That early version looks different from the research-based map described here. It's a circle that stands for the person, with sections for what the person is thinking, feeling, saying, doing and hearing, and it asks you to "project yourself into that person's experience". In a comment on the post, Dave Gray calls it a way "to make a quick approximation that you can then test". That is still a fair use, as long as you label the result as assumptions.
Some versions have more sections than the four boxes used here. The Interaction Design Foundation's article on empathy mapping says Dave Gray's updated canvas has sections for the user's goals and what they see, hear, desire and struggle with. Those are worth adding when you've researched them. Our template stays with the four boxes, which are enough for a round of interviews.
Frequently asked questions
What are the four parts of an empathy map?
Says, Thinks, Does and Feels. Says holds direct quotes, Thinks holds what goes through the person's mind, Does holds their behaviour, and Feels holds their emotions.
What is an empathy map used for?
It gives a team a shared picture of one user or user type, based on research. It's most useful right after interviews, to spot where what people say and what they do disagree, and to decide what to ask next.
Can you make an empathy map without user research?
You can, but it's a list of guesses, so label it that way. A map from assumptions can show a team how much it doesn't know, and it should be replaced once you've spoken to users.
What is the difference between an empathy map and a persona?
An empathy map records what you've learned about a person's words, thoughts, actions and feelings. A persona is a profile of a typical user, with a name, a role, goals and frustrations. An empathy map can be an input to a persona.
How many empathy maps do you need?
One per user or user type. If two groups of users behave differently, combining them in one map hides the difference, so make two.
Map your own user
Start from your last round of interviews while they're still fresh, and keep a transcript open as you write. Open the empathy map template, replace the home buyer's notes with your user's, and share the board link with whoever sat in on the interviews. When the map is done, write down the contradictions, and take them into the next conversation.