Empathy map
Organize what one person sees, hears, thinks and feels, says and does, struggles with and hopes for, keeping records distinct from assumptions.
Use an empathy map when a team needs to bring interview and observation material together around one customer task. This six-part version helps expose research gaps. Compared with a persona, it concentrates on experience in a particular situation rather than adding a biography.
One defined person and task, numbered interview quotes and observation records, the scope of those records, and assumptions still to investigate.
A six-part empathy map with evidence types and specific questions for the next interview or observation.
On this page
How to use it
- Define one person, task and research scope; do not combine different people or situations.
- Number interview and observation records, preserving the exact quote or recorded action first.
- Organize the six areas: see, hear, think and feel, say and do, pains, and gains. Label each entry as a quote, self-report, observation or hypothesis.
- Keep contradictions and gaps visible. Check whether thoughts and feelings were expressed by the person or inferred by the team.
- Choose a follow-up research question and update the map after new evidence. A completed map does not validate a product requirement.
Keep its limits in mind
Behavior does not reveal a person's thoughts automatically. Label quotes, observations, self-reports and team interpretations separately. One person's records do not establish a pattern across all customers, and a self-report is not observed behavior. Leave unsupported areas unknown rather than filling every space.
Understanding one lunch customer's question about the total price
Fictional teaching example: Lantern Lunch has one interview, I1, and one ordering observation, O1, for participant U1 in the same situation. I1 includes 'I want to know the final total before placing my order.' O1 records a return to the basket to read delivery information. These are not records for a customer segment.
On a small screen, scroll across the diagram to read it.
Preserve the record: The colleague's statement is U1's report, not a verified delivery policy. The observed return to the basket supplies no direct evidence of what U1 felt.
Work through the six areas: Place the quote and action in their relevant areas. Mark 'concern about unclear charges' as an interpretation, and preserve the unknown reason for returning and the unknown order outcome.
Result: The map replaces 'all customers think it is too expensive' with traceable entries and research questions. It establishes neither a segment-wide need nor the effectiveness of a redesigned price display.
I1 and O1 concern only U1 in this teaching scenario. The six-part map is not a psychological diagnosis, and the prototype observation has not taken place.
Next step: The researcher should first ask U1 why they returned, then confirm whether a prototype observation can be arranged and record the actual feedback.
Three reasons to leave a map entry unresolved
First, pause when the subject or situation changes. A person ordering lunch alone and a coordinator ordering for a team may face different constraints. Keep the participant, task, and source window attached to each entry. One interview and one observation of the same person are two records, not two independent customers.
Second, do not infer an emotion from an action without checking. Returning to a basket can have several explanations. It does not automatically mean surprise, distrust, or a refusal to pay. Record what was seen, what the person said, and the team’s proposed interpretation separately. An empty thoughts area is more useful than an invented diagnosis.
Third, stop when a map is being used to establish how common a need is or to approve a design. A traceable single-person case can suggest a research question; it does not estimate a segment’s preferences. A hoped-for gain is also different from a measured outcome after redesign. Ask what further evidence the next decision actually requires.
Choose experience evidence, a profile, a journey, or a concept
This guide uses six areas: see, hear, think and feel, say and do, pains, and gains. Layouts vary. The important discipline is keeping the evidence and its status available, especially when the same record informs several areas. Repeating a quote in two cells does not make it two observations.
| Input and question | Method | Output and boundary |
|---|---|---|
| Interview and observation concern one situation | Empathy map | Traceable entries and interpretations; no mind reading or population estimate. |
| Evidence suggests a recurring pattern across relevant people | Persona | An evidence-based working profile; no invented biography. |
| Customer actions span several touchpoints over time | Customer journey map | Stages, evidence, and gaps; no assumed universal path. |
| A possible response needs a mechanism and a test | Idea concept sheet | A bounded concept; no established demand or effectiveness. |
Map U1’s price question without turning it into a price verdict
The complete fictional Lantern Lunch packet contains interview I1 and ordering observation O1 for the same participant U1 in one lunch-ordering situation. In I1, U1 wants to know the final total before ordering and the arrival window. U1 also reports a colleague saying weekend delivery fees may differ; that policy claim is unverified. O1 records a delivery fee at checkout and U1 returning to the basket to read delivery information. The reason for returning and whether the order was abandoned are unknown.
Start with the two records, not the conclusion that all customers consider the service expensive. I1 supports a stated wish for earlier clarity. O1 supports a recorded navigation action. They can be placed together to identify a question, but the observed action does not verify the person’s motive. The colleague’s statement is hearsay reported by U1, not a second participant interview or a confirmed pricing rule.
The table shows the entire six-area map with those boundaries. “Concern about unclear fees” is a team interpretation that could be checked with U1. No fee amount, comparison price, delivery policy, or completed-order result has been supplied. Do not fill those gaps to make the map look complete.
| Area | Record and status | Boundary or next check |
|---|---|---|
| See | O1: fee at checkout; delivery information in basket | Observed display in this situation, not every screen or date. |
| Hear | I1 self-report: colleague says weekend fees may differ | Unverified report; check an authoritative current policy if relevant. |
| Think and feel | I1 wants earlier clarity; team hypothesizes concern | Ask U1 rather than treat the hypothesis as a thought transcript. |
| Say and do | I1 states the wish; O1 records return to basket | Different evidence types from one participant. |
| Pains | I1: difficulty confirming the total early | No evidence that price is high or the order was abandoned. |
| Gains | I1: total including delivery and arrival window | Desired information, not a successful redesign outcome. |
Check an interpretation before expanding the claim
The conclusion is a more precise research question: why did U1 return, and what information was needed at that point? First ask U1 using the relevant event without suggesting the expected answer. Retain the actual explanation, including disagreement with the team’s hypothesis. Participation and a further observation are not confirmed in this case. Do not write a plausible answer into the map while waiting.
If a total-price prototype is later arranged, agree the task, fee assumptions, arrival-window meaning, and what the prototype can demonstrate. Ask the participant to identify the total and any remaining conditions; observe the action separately from what they say about it. A correct answer on a prototype does not establish real checkout accuracy, willingness to pay, or fewer abandoned orders.
At review, keep the original I1 and O1 entries and add the new source with its own status. Revise the interpretation if needed, and list what still requires broader research. Count participants distinctly from sessions and from statements. A repeated statement by the same person cannot establish prevalence across customers. No segment-wide demand, fee problem, or new-design result is reported here.
Check your work
- Each area contains evidence or an explicit unknown, with the evidence type traceable.
- One participant's statements and behavior have not been generalized to a customer segment.
- Feelings and pains have not become proven causes; the next research question is specific.
Framework references
References explain the framework and its scope; they do not validate the fictional example's results.
Put the framework to work
Copy the outline and add your question, evidence, and open questions.