Empathy mapping, Think Say Feel Do
A structured qualitative methodology that generates inputs into an Empathy Map — capturing what people think, say, do and feel about a brand through their own words. Rather than asking participants to evaluate the brand, it surfaces the lived experience behind it: the associations, behaviours, emotions, and moments of use that define the relationship.
The qualitative theory
Section titled “The qualitative theory”Empathy Mapping is a well-established design and strategy tool that organises consumer insight into four quadrants: what people Think (beliefs and associations), Say (expressed opinions and recommendations), Do (behaviours and usage patterns), and Feel (emotional responses). The framework is powerful because it separates these dimensions — which are often conflated in standard research — and makes the structure of the insight visible.
This methodology is designed to generate clean inputs into each of those quadrants through a single conversation. The laddering sequence mirrors the structure of the map: it starts with immediate associations (Think), surfaces expressed opinions (Say), explores behavioural patterns (Do), and then goes deep into the emotional dimension (Feel) — closing with a question that asks participants to locate the source of those feelings.
The result is not just insight about the brand, but insight organised by type — making it directly usable in Empathy Mapping sessions without requiring additional interpretation or re-coding.
The five-question laddering arc is sequential and deliberate. Each question addresses a different quadrant of the map, and together they build from the cognitive (what comes to mind) through the social (what you’d say) and behavioural (what you do) to the emotional (how it feels and why). This progression mirrors how people naturally move from surface-level brand awareness toward deeper, more personal experience — each question builds on what came before.
Configured settings
Section titled “Configured settings”This section documents every setting in the methodology so you can understand the intent behind each choice.
Research framing
Section titled “Research framing”| Field | Value |
|---|---|
| Context & Background | Inputs into Empathy Mapping to capture what people think, say, do and feel about a brand. |
| Research Objectives | Understand how consumers perceive and experience the brand through their own thoughts, words, actions and emotions. Identify key needs, pain points, motivations and emotional drivers. |
| AI Moderator Tone | Friendly and inquisitive |
The tone here is friendly and inquisitive rather than the casual register used in most other brand methodologies. Inquisitive signals genuine curiosity — it tells participants the AI Moderator wants to understand, not just record. This is well suited to an empathy-focused study where the goal is depth of understanding rather than breadth of coverage.
Conversation introduction
Section titled “Conversation introduction”Hi, I’m
{{companionName}}, your survey companion today. I’m here to learn about your views and experiences with BRAND. I’ll ask a few short questions, and of course, there are no right or wrong answers, we’re just interested in your honest perspective.When BRAND comes to mind, the first things that come to mind are…
The introduction does two things. First, it frames the conversation as a listening exercise — “I’m here to learn about your views” — which sets an empathetic, participant-centred register from the start. Second, the opening prompt ends as an incomplete sentence: “the first things that come to mind are…” This invites participants to complete the thought in their own words rather than respond to a direct question, which produces more immediate, unfiltered associations.
Laddering questions
Section titled “Laddering questions”Five questions, with Randomise off and Paraphrase off:
| # | Question | Empathy map quadrant | Purpose |
|---|---|---|---|
| 1 | What makes you think of that? | Think | Probes the reasoning or association behind the opening response — surfaces the belief or connection that produced the first impression |
| 2 | If someone asked your honest opinion of BRAND, what would you say? | Say | Captures what participants would express publicly — the words, recommendations, or criticisms they’d actually voice to others |
| 3 | Describe how you typically use BRAND in as much detail as possible | Do | Surfaces behavioural patterns — the routines, occasions, and contexts of actual use |
| 4 | How does using BRAND usually leave you feeling — describe that feeling in detail | Feel | Shifts to the emotional dimension — asks for the affective response to the brand experience, with a deliberate push for specificity (“in detail”) |
| 5 | What is it about BRAND that leaves you feeling like that? | Feel | Grounds the emotion in its cause — connecting the feeling back to something concrete about the brand or the experience of using it |
Randomise is off because these questions follow the structure of the Empathy Map itself — Think → Say → Do → Feel. Disrupting the order would break the quadrant-by-quadrant logic that makes the outputs directly usable in mapping sessions.
Paraphrase is off for the same reason. The specific wording of each question is matched to its quadrant: Q2 uses “honest opinion” to unlock the Say quadrant; Q3 uses “in as much detail as possible” to generate rich Do data; Q4 uses “usually leave you feeling” to access the affective layer rather than a one-off emotional reaction. These phrasings are intentional and should not vary.
Questions 4 and 5 work as a pair. Q4 surfaces the emotion; Q5 grounds it in the experience that produced it. This two-step sequence produces emotionally rich data that remains anchored in something observable, which is more useful for Empathy Mapping than free-floating sentiment.
Default probe
Section titled “Default probe”Tell me more about
{{topic}}?
Paraphrase on. A simple, open probe that gives the AI Moderator room to follow up on any of the four core quadrant topics without steering the direction of the response.
Default prompt
Section titled “Default prompt”What about
{{topic}}, can you tell me more about that?
Paraphrase off. The prompt is kept consistent across participants so that when a topic is introduced that the participant hasn’t mentioned, it’s introduced in the same way each time. This is particularly important for the Trigger to use topic, which has a custom prompt override (see Topics below).
Probe & prompt limit
Section titled “Probe & prompt limit”9 — with four topics carrying one probe each and one topic with a prompt, the practical ceiling is well below 9. The higher limit provides headroom for the default probe to fire naturally during the laddering phase without hitting the cap before all topics have been explored.
Topics
Section titled “Topics”| Topic | Description | Probes | Prompts |
|---|---|---|---|
| Think | What comes to mind about the brand, including beliefs, associations, and perceptions | 1 | Off |
| Say | How people would talk about the brand out loud, including praise, criticism, and recommendations | 1 | Off |
| Do | How people typically use the brand in real life, including routines, occasions, and behaviours | 1 | Off |
| Feel | How the brand leaves people feeling, beyond generic positive reactions, and the emotions it evokes | 1 | Off |
| Trigger to use | What prompts or triggers someone to use the brand | Off | On |
Randomise probe order: off. Randomise prompt order: off.
The four core Empathy Map quadrant topics (Think, Say, Do, Feel) all have probing enabled but prompting disabled. This is deliberate — the laddering questions are specifically designed to cover each quadrant, so prompting would introduce redundancy and risk disrupting the flow of the structured sequence. The probes exist to go deeper when a quadrant topic is raised spontaneously outside the laddering arc.
Trigger to use inverts this pattern: probing off, prompting on. Triggers — the moments that prompt someone to reach for a brand — are not a standard Empathy Map quadrant, but they are high-value insight that doesn’t reliably surface on its own. The prompt ensures the topic is introduced if the participant hasn’t mentioned it. The probe is off because once a trigger is described, the laddering and default probe provide enough follow-up.
The prompt override for Trigger to use is: What typically prompts you to use this brand?
This phrasing is more specific than the default prompt and more likely to produce a usable behavioural trigger rather than a general attitudinal response.
Conversation ending
Section titled “Conversation ending”- Closing question: Finally, if you had to describe your relationship with BRAND in just a few words, what would you say?
- Dynamic ending: On
The closing question invites participants to synthesise everything they’ve described — associations, opinions, behaviours, and emotions — into a compact, natural formulation. “Your relationship with BRAND” frames it as a personal connection rather than a brand evaluation, which produces more candid and emotionally resonant closing statements.
Dynamic ending is on so the AI Moderator can close in a way that reflects the depth and tone of the individual conversation.
Adapting this methodology
Section titled “Adapting this methodology”When you apply this methodology to a new conversation, these are the fields you’ll typically need to change:
- Replace BRAND in the conversation introduction, all five laddering questions, and the closing question with your brand name.
- Context & Background — add context about the specific brand, category, and participant group.
- Research Objectives — update to reflect which quadrants or dimensions are most important for your project. If the study is primarily about emotional drivers, you may want to add probing to the Feel topic. If it’s about usage behaviour, consider increasing the probe count on Do.
- Topics — the five defaults map directly to the Empathy Map framework. Add category-specific topics if your mapping sessions require additional dimensions. Keep descriptions grounded in what participants would actually say, not in abstract research constructs.
- Probe & Prompt Limit — scale up if you add more topics with active probing or prompting.
The laddering sequence — and particularly the paired structure of questions 4 and 5 — should be left unchanged wherever possible. The sequence is the methodology’s core mechanism, and it is designed to produce outputs that slot directly into each Empathy Map quadrant without additional recoding.
Frequently asked questions
Section titled “Frequently asked questions”How does this methodology connect to Empathy Mapping?
Each laddering question is designed to generate inputs into a specific Empathy Map quadrant. Q1 (what makes you think of that?) → Think. Q2 (honest opinion you’d share) → Say. Q3 (how you typically use it) → Do. Q4 and Q5 (how it leaves you feeling, and why) → Feel. The conversation produces structured, quadrant-labelled data that can be fed directly into an Empathy Mapping session without needing to re-sort or re-interpret the outputs.
Why is the AI Moderator Tone “friendly and inquisitive” rather than “friendly and casual”?
Empathy mapping is fundamentally about understanding — the goal is to get inside the participant’s experience of the brand, not just to record their opinions. An inquisitive register signals genuine curiosity and invites more reflective, considered responses than a casual tone does. For research that will feed into persona development or strategy workshops, depth of understanding matters more than conversational ease.
Why are prompts disabled on Think, Say, Do, and Feel?
The laddering questions are specifically designed to elicit content for each of these quadrants in sequence. Prompting them would introduce redundancy — participants have already been asked directly about each dimension. If a topic is mentioned outside the laddering arc, the probe provides follow-up. Prompting is reserved for Trigger to use, which the laddering doesn’t cover.
Why is Paraphrase off for the Default Prompt but on for the Default Probe?
Probes are conversational follow-ups where natural variation is appropriate and even desirable — they should feel like a genuine response to what the participant said. Prompts introduce a topic the participant hasn’t mentioned, and consistency matters more here: the Trigger to use prompt in particular has a specific wording that is more likely to produce a useful behavioural response. Paraphrase off ensures that wording is used consistently across the sample.
Can I use this methodology if I’m not running a formal Empathy Mapping session?
Yes. The Think / Say / Do / Feel framework is a useful analytical lens regardless of whether you’re running a formal mapping session. The structured outputs make it easy to compare responses across participants and identify patterns within each dimension. If you’re not using Empathy Mapping specifically, you can treat the four quadrants as four distinct analytical categories and analyse them independently.