Empathy, Customer Discovery and Design Research

Introduction

Entrepreneurial opportunities rarely begin with a perfectly defined problem. They emerge through disciplined engagement with people, organisations and contexts in which needs are experienced. Design research helps an entrepreneurial team understand what customers do, why they do it, what they struggle with, and what outcomes they value. It is therefore different from merely asking people whether they like an idea. Its purpose is to generate credible insight before committing substantial resources to a product, service or business model.

This chapter presents a practical and critical approach to empathy, customer discovery and design research. It focuses on interviews, observation, contextual inquiry, diary studies and digital research. It also explains how to identify customer jobs, behaviours, motivations, pain points and unmet needs while minimising researcher bias. Because research involves real people and potentially sensitive information, ethical conduct is treated as a central design requirement rather than an administrative formality.

Learning Objectives

By the end of this chapter, you should be able to:

1. Explain empathy as a disciplined research capability rather than simple sympathy.
2. Distinguish customer discovery from market validation and usability testing.
3. Select appropriate qualitative and digital research methods.
4. Plan and conduct ethical interviews, observations, contextual inquiries and diary studies.
5. Analyse evidence to identify jobs, behaviours, motivations, pain points and unmet needs.
6. Recognise and reduce common forms of researcher and participant bias.
7. Translate research findings into opportunity statements and testable assumptions.
8. Judge the quality, limitations and ethical implications of design research.

1. Empathy as an Entrepreneurial Capability

Empathy in design is the effort to understand another person’s situation, goals, feelings, constraints and interpretations without assuming that one’s own experience is representative. It is not agreement with every customer statement, nor is it an invitation to stereotype users. Empathy is an evidence-seeking practice that combines attentive listening, observation and interpretation.

Entrepreneurs need empathy for three reasons. First, customers often describe solutions rather than underlying needs. A person may request a mobile application when the deeper need is reassurance, coordination or reduced administrative effort. Second, behaviour is frequently inconsistent with stated preference. People may claim that convenience is most important but repeatedly choose a slower option because it feels safer or more trustworthy. Third, needs are situated: the same person may behave differently at home, at work, under time pressure or with other people present.

A useful distinction is between what people say, do, use and feel. Interviews are especially useful for meanings, memories and aspirations; observation reveals practices and workarounds; artefacts reveal tools and constraints; diaries reveal change over time. Strong research triangulates these forms of evidence rather than treating any one as complete.

Empathy should also be bounded by respect. Researchers must not treat people as sources of quotations or unpaid product consultants. Participants contribute time, knowledge and sometimes vulnerability. Ethical research therefore requires informed consent, privacy protection, fair treatment and responsible use of findings.

2. Customer Discovery and Design Research

Customer discovery is the systematic investigation of a target situation before, or alongside, the development of a proposed offering. It asks questions such as:

• Who experiences the problem?
• In what circumstances does it arise?
• What are people trying to accomplish?
• How do they currently cope?
• What costs, risks or frustrations result?
• Who influences, pays for, delivers or approves the solution?

Design research is broader than customer discovery. It may include research with customers, employees, partners, regulators and communities. Customer discovery is generally opportunity-oriented; design research may also support service design, product improvement, organisational change and social innovation.

Customer discovery should not be confused with:

A central principle is to study the problem space before over-investing in the solution space. Showing a polished concept too early can make participants polite, anchor their responses and encourage the researcher to defend the idea. Early research should usually explore experiences and existing behaviour before presenting a proposed solution.

3. A Framework for What to Learn

A practical research framework organises evidence into six connected categories.

3.1 Customer jobs

A job is the progress a person is trying to make in a particular situation. Jobs may be:

• Functional: completing a concrete task, such as submitting a claim.
• Emotional: feeling confident, safe or in control.
• Social: appearing competent, responsible or accepted.

A job is not necessarily a product feature. “Use a budgeting app” is a solution activity; “make a reliable spending decision before payday” is closer to a job.

3.2 Behaviours

Behaviours are observable actions, sequences, choices and workarounds. Researchers should record what happened, when, with what tools, and under what constraints. “The participant is disorganised” is an interpretation; “the participant copied figures from three messages into a notebook before making the booking” is an observation.

3.3 Motivations

Motivations include desired outcomes, values, fears, incentives and perceived trade-offs. They may be explicit or inferred. Inference must remain provisional until supported by further evidence.

3.4 Pain points

Pain points are experienced difficulties, including delays, errors, uncertainty, financial cost, effort, anxiety and loss of control. Severity depends not only on frequency but also on consequence and the availability of alternatives.

3.5 Unmet needs

An unmet need exists when an important desired outcome is inadequately served by current options. A complaint alone does not establish an unmet need. Researchers should investigate whether the problem is important, recurring, underserved and associated with willingness or ability to change.

3.6 Context and stakeholders

The direct user may not be the buyer, beneficiary, decision-maker or gatekeeper. A workplace software product, for example, may involve employees, managers, procurement officers, information-security staff and customers. Mapping these relationships prevents the research from representing only the easiest participant to recruit.

A concise opportunity statement can combine these elements:

When [situation] occurs, [customer] is trying to [job], but [barrier or pain] makes it difficult, resulting in [consequence].

This statement is an interpretation, not a proven fact. It should be traceable to specific evidence.

4. Choosing a Research Method

No single method reveals the whole customer experience. Selection should follow the research question, not researcher preference.

Interviews

Interviews are directed conversations used to explore experiences, meanings, decisions and perceptions. They are appropriate when the researcher needs to understand events that cannot be observed directly, such as a past decision or private concern.

Observation

Observation records behaviour in a natural or realistic setting. It is useful for discovering routines, interruptions, physical constraints and workarounds that participants may forget or consider too ordinary to mention.

Contextual inquiry

Contextual inquiry combines observation with brief questions during the activity. The participant remains the expert in the context, while the researcher asks for clarification at relevant moments. It is particularly valuable for complex work, service processes and tacit knowledge.

Diary studies

A diary study asks participants to record experiences repeatedly over a period. Entries may include text, photographs, audio or structured prompts. Diary studies are useful for episodic, variable or emotionally sensitive experiences, such as commuting, caregiving, symptom management or purchasing decisions.

Digital research

Digital research can include analysis of search behaviour, online communities, app-store reviews, support tickets, social-media discussions, clickstream data and remote interviews. Digital traces may reveal scale and naturally occurring language, but they are not automatically representative. Algorithms, platform cultures and self-selection shape what becomes visible.

Method choice should consider sensitivity, access, participant burden, safety, time and the consequences of being wrong.

5. Planning Ethical Research

A research plan should specify the decision the research will inform, the uncertainties to investigate, the participants required, the method, recruitment process, analysis approach, risks and safeguards.

Informed consent

Consent should be informed, voluntary, specific and revocable. Participants should understand:

• the purpose of the study;
• what participation involves and how long it will take;
• whether audio, video, photographs or screen recordings will be collected;
• foreseeable risks and benefits;
• how data will be stored, used and reported;
• who may access the information; and
• how to withdraw, where withdrawal is practicable.

Consent is a process, not merely a signed form. Researchers should check understanding, especially when working with children, people with limited language proficiency, employees, patients or economically dependent participants.

Privacy and confidentiality

Researchers should collect only data necessary for the research purpose. Names, contact details, account identifiers, health information, financial information and location data require careful handling. Pseudonymisation replaces direct identifiers but does not guarantee anonymity; combinations of details can re-identify a person. Access should be limited, files encrypted where appropriate, retention periods defined, and quotations reviewed for deductive-identification risk.

Power and vulnerability

Employees may feel unable to refuse a manager’s request. Customers may fear losing service. Researchers should make clear that participation will not affect employment, benefits or eligibility. Incentives should compensate time without becoming coercive. Sensitive topics require appropriate referral or escalation procedures; researchers should not promise support they cannot provide.

Digital ethics

Publicly visible online content is not automatically ethically free to reproduce. Users may not expect their posts to be analysed, quoted or linked to their identities. Researchers should consider platform terms, reasonable expectations of privacy, community norms, consent feasibility and the sensitivity of the topic. Automated scraping and artificial-intelligence analysis must comply with applicable law, institutional policy and the original purpose of collection.

6. Conducting High-Quality Interviews

A good interview is neither an interrogation nor a sales presentation. It is a structured conversation in which the researcher creates conditions for accurate, detailed accounts.

Interview guide

An interview guide normally contains an opening, warm-up, core topics, probes and closing. Questions should move from broad experience to specific episodes. The guide should be flexible enough to follow unexpected but relevant evidence.

Weak question: “Would you use an app that automatically solves this problem?”

Better questions: “Tell me about the last time this happened.” “What did you do first?” “What made that difficult?” “What alternatives did you consider?” “What happened next?”

The first question asks for a prediction and invites politeness. The latter questions seek behaviour and context.

Interviewing techniques

Use open questions, neutral probes and deliberate pauses. Ask one question at a time. Request examples, sequences and artefacts. Distinguish a participant’s direct statement from the researcher’s interpretation. Avoid leading language, jargon, false familiarity and excessive self-disclosure. When a participant makes a general claim, ask for the most recent instance.

A useful sequence is:

1. Establish context and consent.
2. Ask about the participant’s role and routine.
3. Explore a recent concrete episode.
4. Examine actions, tools, decisions and consequences.
5. Investigate workarounds and alternatives.
6. Explore emotional and social dimensions.
7. Ask what would make the situation meaningfully better.
8. Summarise understanding and invite correction.

Recording improves accuracy but requires explicit permission. If recording is not permitted, two researchers may divide conversation and note-taking, provided their presence does not inhibit participation.

7. Observation and Contextual Inquiry

Observation should be planned as systematic attention, not passive watching. Before entering the setting, define what is relevant: sequence, time, actors, tools, interruptions, handoffs, exceptions and visible outcomes. Record descriptive notes separately from interpretations.

A field-note format may contain:

Contextual inquiry uses short questions without unnecessarily interrupting the task. Appropriate probes include “What are you checking here?”, “What would happen if this option were unavailable?” and “How do you know this is complete?” Researchers should not correct the participant or redesign the process during observation.

The observer effect cannot be eliminated entirely. It can be reduced by spending sufficient time in the setting, behaving unobtrusively, acknowledging the effect of observation and comparing observed behaviour with accounts from other sources. Safety, confidentiality and permission from all relevant parties must be considered, particularly in workplaces, healthcare settings and private homes.

8. Diary Studies and Digital Research

Diary studies are valuable when memory is weak or experiences unfold over time. A sound diary protocol defines the event that should trigger an entry, the expected response time, the duration, examples of useful detail and a mechanism for support. Prompts should be short enough to sustain participation.

For example, after each public-transport disruption, a participant might record the time, intended destination, action taken, emotional response and outcome. Researchers should not infer that missing entries mean no events occurred. Compliance patterns are themselves data but may reflect burden, access or privacy concerns.

Digital research requires attention to provenance. Researchers should document where data came from, when it was collected, what was excluded, how duplicates and automated content were treated, and whether the dataset represents users, vocal minorities or platform-specific populations. Sentiment analysis and large-language-model summarisation can accelerate coding, but outputs may reproduce demographic, linguistic or cultural bias. Human review, audit samples and clear uncertainty statements remain necessary.

9. Sampling and Research Quality

Qualitative research generally seeks information-rich variation rather than statistical representation. A purposive sample may include people with different levels of experience, contexts, access conditions and roles. Snowball recruitment can reach hidden populations but may produce socially connected samples. Convenience sampling is efficient but often overrepresents accessible and enthusiastic participants.

Researchers should define inclusion and exclusion criteria and explain why each segment matters. Sample size should be guided by the research purpose, diversity of contexts and diminishing emergence of materially new insights—not by a universal numerical rule. Repeated themes alone do not prove prevalence.

Quality can be improved through:

• Triangulation: comparing interviews, observations, artefacts and digital traces.
• Reflexivity: recording how the researcher’s position may shape access and interpretation.
• Member reflection: inviting participants to correct factual misunderstandings, while recognising that agreement does not prove an interpretation.
• Peer debriefing: asking another researcher to challenge conclusions.
• Negative-case analysis: actively examining evidence that contradicts the emerging pattern.
• Auditability: retaining a clear link between claims, evidence and analytical decisions.

Research quality is not achieved by pretending to be completely objective. It is achieved by making the pathway from evidence to interpretation visible and contestable.

10. Analysing Research Data

Analysis begins during collection. Researchers should write brief reflections after each session, identify new questions and revise the guide without losing comparability.

A practical analysis process is:

1. Familiarise: review notes, recordings or artefacts repeatedly.
2. Segment: identify meaningful statements, actions or episodes.
3. Code: label content such as “manual workaround,” “fear of error” or “approval dependency.”
4. Compare: examine similarities and differences across people, contexts and time.
5. Cluster: group related codes into themes.
6. Interpret: connect themes to jobs, behaviours, motivations, pains and outcomes.
7. Challenge: search for disconfirming evidence and alternative explanations.
8. Synthesize: create evidence-based opportunity statements and assumptions.

A code should describe the content without prematurely embedding a solution. “Needs a dashboard” is a solution code; “cannot see status across channels” describes a research finding.

An insight is stronger when it includes context, behaviour, meaning and consequence:

During end-of-month reporting, small-business owners reconcile information across messaging and spreadsheet tools, because no single source is trusted; this creates uncertainty and delays decisions.

The claim should be linked to source evidence and labelled appropriately: observed, directly reported, inferred or still uncertain.

11. Minimising Researcher Bias

Bias is not limited to intentional prejudice. It can enter through the research question, recruitment, wording, setting, analysis and reporting.

Reflexive practice includes maintaining a research journal, stating assumptions before fieldwork, rotating interview roles, separating observation from interpretation, and seeking critique from someone who does not share the project’s commercial commitment. Teams should also examine whose voice is absent and who may bear the costs of a proposed innovation.

12. From Evidence to Opportunity

Research does not automatically produce a business opportunity. Teams must assess whether a discovered problem is important, persistent, underserved and connected to a feasible value proposition. A useful opportunity assessment considers:

• Desirability: Is the outcome valuable to a meaningful group?
• Feasibility: Can the organisation deliver it technically and operationally?
• Viability: Can value and revenue or impact be sustained?
• Ethical acceptability: Does the approach respect rights, safety and fairness?

A customer insight should generate a testable assumption rather than a conclusion such as “customers love our idea.” For example:

We believe independent tutors need a faster way to reconcile payments because they currently spend time comparing messages and bank records. We will test this by observing reconciliation, measuring completion effort and examining whether tutors adopt a low-fidelity alternative.

The assumption identifies a customer, context, behaviour and evidence requirement. Research continues as the idea becomes more specific.

13. Illustrative Case: Discovering the Real Problem in Medication Support

This is a fictional illustrative case.

Background

A student venture proposed a mobile reminder application for adults managing several prescriptions. The team assumed that missed doses were primarily a memory problem.

Problem and initial approach

The founders interviewed six people using the question, “Would automatic reminders help you take medication on time?” Most participants responded positively. The team initially interpreted this as strong demand.

A design researcher challenged the finding because the question described a solution and did not reveal current behaviour. The team redesigned the study. They recruited participants with variation in age, medication complexity and living arrangements, obtained consent, and conducted interviews alongside a medication-routine diary. They did not collect prescription names or medical details unless necessary.

Analysis

Participants did miss doses, but the episodes differed. Some deliberately postponed medication because of side-effect concerns. Others were unsure whether they had already taken it. One participant relied on a family member because instructions changed frequently. A reminder alone could increase anxiety or lead to unsafe duplication. Observations also showed that medication was stored in different locations, and participants used calendars, packaging and informal family messages.

The deeper job was not simply “remember to take medication.” It was “make a safe, confident decision about medication at the right time despite changing routines and uncertainty.” The team therefore identified multiple stakeholders, including clinicians, caregivers and pharmacists.

Approach and solution direction

Rather than immediately building an alarm, the team tested a paper-based routine guide and a low-fidelity confirmation flow. Participants were asked to explain what information they needed when uncertain. The prototype was not presented as medical advice. The team planned clinical review, clear escalation guidance and strict data minimisation before any digital development.

Outcome

The research did not validate the original reminder concept. It prevented the team from building a potentially unsafe feature and redirected attention toward coordination, clarity and escalation. Further research was required before making health claims or deploying the service.

Learning points

1. Positive reactions to a solution do not establish a need or safe use case.
2. Concrete episodes reveal distinctions hidden by broad questions.
3. The user, buyer, beneficiary and safety authority may be different actors.
4. Ethical constraints can improve the quality of the opportunity, not merely restrict innovation.
5. In high-risk domains, uncertainty must be reported rather than disguised as insight.

14. Challenges, Limitations and Contemporary Practice

Design research is vulnerable to access problems, small samples, incomplete records, participant fatigue, commercial pressure and overconfident synthesis. Qualitative depth does not provide prevalence estimates, and digital scale does not guarantee validity. Researchers should avoid claiming that “users” in general behave like a few participants.

Artificial intelligence can assist transcription, translation, clustering and search, but it can also remove hesitation, flatten cultural meaning, infer sensitive attributes and produce plausible but unsupported summaries. Organisations should define approved tools, disclose automated processing where appropriate, protect confidential data, validate outputs against source material and retain human accountability for decisions.

Contemporary research increasingly requires inclusive design. Recruitment should consider disability, language, connectivity, age, gender, cultural context and economic conditions. Accessibility is not merely a compliance issue: excluding people can cause teams to misidentify the problem and create products that work only for privileged users.

The best practice is a proportionate research cycle: clarify the decision, investigate the context, gather varied evidence, analyse transparently, challenge assumptions, protect participants and convert findings into the next test. Research is not a ceremonial phase before “real” innovation. It is the mechanism through which entrepreneurial action becomes more informed, responsible and relevant.

Chapter Summary

Empathy and customer discovery are disciplined ways of learning how people experience situations and pursue desired progress. Interviews reveal accounts and meanings; observation and contextual inquiry reveal situated behaviour; diary studies reveal change over time; digital research can provide naturally occurring language and large-scale signals but requires careful interpretation. Customer jobs, behaviours, motivations, pain points, unmet needs and stakeholder relationships provide a structure for analysis.

Ethical research requires informed and ongoing consent, privacy protection, proportionate data collection, attention to power and vulnerability, and responsible digital practices. High-quality research uses concrete episodes, neutral questions, purposive variation, triangulation, reflexivity, negative-case analysis and an auditable link between evidence and claims. Researchers must distinguish observations from interpretations and insights from validated market opportunities. The ultimate output is not a collection of attractive quotations but a set of defensible, testable assumptions that can guide responsible entrepreneurial experimentation.

Key Terms

Empathy: Disciplined effort to understand another person’s situation, experience and perspective.

Customer discovery: Systematic investigation of customer situations, needs, behaviours and alternatives.

Design research: Research conducted to understand people, contexts, systems and opportunities for design.

Customer job: Progress a person is trying to make in a particular situation.

Pain point: Difficulty, cost, risk, uncertainty or frustration experienced during an activity.

Unmet need: An important desired outcome that current alternatives inadequately serve.

Contextual inquiry: Observation combined with brief, in-context questioning.

Diary study: Repeated participant-recorded documentation of experiences over time.

Triangulation: Comparison of evidence from different methods, sources or researchers.

Informed consent: Voluntary agreement based on adequate understanding of a study and its implications.

Pseudonymisation: Replacing direct identifiers while recognising that re-identification may remain possible.

Purposive sampling: Deliberate selection of participants for relevant characteristics or variation.

Researcher bias: Systematic influence of researcher assumptions or procedures on evidence or interpretation.

Reflexivity: Critical examination of how the researcher shapes the research process.

Negative-case analysis: Deliberate examination of evidence that contradicts an emerging pattern.

Workaround: An improvised method used to cope with limitations in an existing process or system.

Opportunity statement: Evidence-based description of a situation, job, barrier and consequence.

Desirability: The extent to which an offering addresses a valued outcome.

Member reflection: Inviting participants to clarify or challenge a researcher’s account.

Data minimisation: Collecting only information necessary for a defined purpose.

Review Questions

Short Answer Questions

1. Why should researchers study concrete episodes rather than rely only on general opinions?
2. Distinguish a customer job from a product feature.
3. What is the purpose of contextual inquiry?
4. State two requirements of informed consent.
5. How does negative-case analysis improve research quality?

Long Answer Questions

1. Compare interviews, observation, contextual inquiry and diary studies in terms of purpose, evidence and limitations.
2. Explain how customer discovery differs from concept testing, usability testing and market sizing.
3. Discuss the ethical issues involved in collecting and analysing digital research data.
4. Explain a complete process for analysing qualitative research from raw notes to an opportunity statement.
5. Discuss major forms of researcher bias and evaluate practical methods for reducing them.

Analytical / Application Questions

1. A founder asks users, “Would you pay for an AI assistant that eliminates your administrative work?” Redesign the research plan and provide six neutral questions that investigate the underlying job.
2. A team observes employees bypassing a new workplace system by using spreadsheets and messaging applications. Analyse what additional evidence is needed before declaring this an unmet need.
3. Design an ethical diary study for a recurring customer experience. Specify the participant criteria, trigger for entries, data to collect, consent safeguards and method for interpreting incomplete entries.

Multiple Choice Questions with Answers

Question 1

Which research method is most directly suited to understanding how a task is actually performed in its natural setting?

A. Market sizing

B. Observation

C. Concept testing

D. Financial forecasting

Correct Answer: B. Observation

Question 2

Which statement best describes a customer job?

A. A specific product feature requested by a customer

B. A demographic category used for market segmentation

C. The progress a person is trying to make in a situation

D. The revenue target assigned to a sales team

Correct Answer: C. The progress a person is trying to make in a situation

Question 3

Which interview question is least likely to introduce solution bias?

A. Would you use an app that solves this problem?

B. Do you agree that the current process is inconvenient?

C. How much would you pay for our proposed service?

D. Tell me about the last time this situation occurred.

Correct Answer: D. Tell me about the last time this situation occurred.

Question 4

What is the primary purpose of purposive sampling in qualitative design research?

A. To guarantee statistical representation of the whole population

B. To deliberately include participants relevant to the research question

C. To recruit only participants who support the concept

D. To eliminate the need for ethical review

Correct Answer: B. To deliberately include participants relevant to the research question

Question 5

Which practice best protects participant privacy?

A. Collecting all available information in case it becomes useful

B. Publishing detailed quotations with identifiable context

C. Collecting only data necessary for the defined research purpose

D. Assuming publicly visible online content requires no safeguards

Correct Answer: C. Collecting only data necessary for the defined research purpose

Question 6

A researcher searches only for evidence supporting a preferred product idea. This is an example of:

A. Confirmation bias

B. Recall bias

C. Sampling variation

D. Member reflection

Correct Answer: A. Confirmation bias

Question 7

What does triangulation involve?

A. Replacing qualitative research with a survey

B. Comparing evidence from different sources or methods

C. Asking the same leading question three times

D. Selecting exactly three participants per segment

Correct Answer: B. Comparing evidence from different sources or methods

Question 8

Why can a reminder application be unsafe in some medication situations?

A. Reminders always reduce adherence

B. A reminder may not resolve uncertainty about whether medication was already taken

C. Digital products cannot collect any user feedback

D. Participants never use mobile devices for health purposes

Correct Answer: B. A reminder may not resolve uncertainty about whether medication was already taken

Question 9

Which statement about qualitative research sample size is most accurate?

A. There is one universal sample size that guarantees validity

B. A larger sample automatically removes researcher bias

C. Size should reflect the purpose, variation and emergence of materially new insights

D. Only statistically random samples are acceptable

Correct Answer: C. Size should reflect the purpose, variation and emergence of materially new insights

Question 10

What is the most appropriate role of artificial intelligence in analysing design research?

A. To make final interpretations without human review

B. To replace informed consent when data are online

C. To assist processing while outputs are checked against source evidence

D. To infer sensitive participant attributes automatically

Correct Answer: C. To assist processing while outputs are checked against source evidence