Customer Discovery and Market Research

Introduction

Entrepreneurship begins with uncertainty. Founders rarely know with confidence who will buy, which problem matters most, how customers currently solve it, or whether a market is large enough to support a viable enterprise. Customer discovery and market research provide disciplined ways to reduce that uncertainty before substantial resources are committed.

Customer discovery is the systematic process of learning about customers, their problems, behaviors, motivations, constraints, and willingness to adopt a solution. Market research is the broader process of collecting and interpreting information about customers, industries, competitors, market conditions, and external forces. The two activities overlap, but they are not identical. Customer discovery focuses especially on the problem and the people experiencing it; market research also examines market structure, demand, competition, regulation, and economic feasibility.

Research does not eliminate entrepreneurial risk. Its purpose is to replace unsupported assumptions with evidence, identify important unknowns, and improve the quality and speed of decisions.

Learning Objectives

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

• define customer segments using meaningful behavioral and contextual criteria;
• explain jobs-to-be-done and distinguish functional, emotional, and social jobs;
• design and conduct ethical primary research;
• evaluate secondary research for relevance, quality, and bias;
• estimate market size using TAM, SAM, and SOM;
• analyze competitors and alternatives;
• develop evidence-based personas and customer insights;
• distinguish observations, interpretations, assumptions, and validated findings; and
• use research evidence to prioritize entrepreneurial decisions.

1. From Entrepreneurial Assumptions to Testable Questions

An early business idea usually contains a chain of assumptions:

1. A particular group experiences a meaningful problem.
2. The problem occurs frequently or has serious consequences.
3. Existing alternatives are inadequate or inconvenient.
4. Customers can and will pay for an improved solution.
5. The venture can reach customers economically.
6. The market is sufficiently large and durable.

A weak research process asks, “Do people like my idea?” This encourages politeness and confirmation bias. A stronger process asks questions such as:

• When did you last experience this problem?
• How do you solve it today?
• What does the current solution cost in money, time, or effort?
• What have you tried previously?
• Who participates in the purchase decision?
• What would prevent adoption?

The distinction between a problem interview and a sales pitch is essential. A researcher seeks accurate understanding, including disconfirming evidence. A salesperson seeks persuasion. Confusing the two produces unreliable findings.

Hypotheses and evidence thresholds

A research hypothesis is a provisional statement that can be supported or weakened by evidence. For example: “Independent restaurants with fewer than 20 employees lose substantial revenue because delivery orders are difficult to coordinate.” The statement identifies a population, a problem, and a consequence.

A useful research plan specifies in advance what evidence would change the decision. For example, a team might continue only if at least 8 of 12 carefully selected restaurant operators describe the problem without prompting and at least 4 have recently paid for an imperfect workaround. This is not a universal statistical rule; it is a decision threshold for an exploratory study. The threshold should be set before reviewing results to reduce motivated reasoning.

2. Customer Segmentation

A customer segment is a group whose members share relevant characteristics, needs, behaviors, or responses to a value proposition. Segmentation is not merely describing people by age or location. The most useful segment is one in which customers have sufficiently similar problems and buying conditions that the venture can serve them with a coherent offer.

Common segmentation bases

A segment should be described operationally. “Busy consumers” is vague. “Urban professionals who purchase prepared meals at least three times per week and order after 7 p.m. because of work commitments” is more researchable.

Segmenting the buying system

The user, buyer, payer, decision maker, and beneficiary may be different people. In a school software market, teachers may use a product, a principal may approve it, a procurement officer may contract for it, and a public authority may pay for it. Research must include the roles that influence adoption rather than interviewing only the eventual user.

Segment attractiveness and strategic focus

A segment may be attractive because it has a serious problem, accessible decision makers, adequate purchasing power, growth potential, and limited competitive resistance. However, a large segment is not automatically a good starting point. Early ventures often benefit from a narrowly defined beachhead segment in which the problem is urgent and the venture can obtain learning quickly.

3. Jobs-to-Be-Done

The jobs-to-be-done (JTBD) perspective focuses on what customers are trying to accomplish in a particular situation. Customers do not simply buy products; they “hire” products or services to make progress.

A job may be:

• Functional: completing a practical task, such as transporting goods or preparing a report.
• Emotional: feeling secure, confident, relieved, or in control.
• Social: appearing competent, responsible, innovative, or successful to others.

A useful job statement identifies context, motivation, desired progress, and constraints: “When independent retailers receive unpredictable deliveries, they want to know arrival times accurately so that they can schedule staff without paying for idle labor.”

JTBD differs from a feature list. “Customers want route optimization” describes a proposed solution. “Customers want to reduce uncertainty in delivery scheduling” describes the underlying job and leaves room for multiple solutions.

Forces affecting adoption

Four forces often influence whether customers switch:

• the push of dissatisfaction with the current situation;
• the pull of a promising new solution;
• anxiety about changing behavior or accepting risk; and
• the inertia of existing habits and processes.

A product can fail even when its benefits are attractive if switching costs, uncertainty, training requirements, or contractual barriers are high. Discovery should therefore investigate both desired outcomes and adoption friction.

4. Designing Primary Research

Primary research collects new evidence directly from relevant people or situations. Common methods include interviews, observation, surveys, diary studies, usability tests, experiments, and pilot transactions.

Exploratory interviews

Interviews are particularly useful early because they reveal language, context, workarounds, and unexpected causes. A semi-structured interview uses a consistent topic guide while allowing follow-up questions.

A practical interview sequence is:

1. explain the purpose, voluntary nature, and expected duration;
2. begin with the participant’s context and recent experiences;
3. ask for a specific recent instance rather than general opinions;
4. investigate current behavior, alternatives, costs, and consequences;
5. explore decision criteria and barriers;
6. ask what the researcher may have misunderstood; and
7. close by explaining next steps and obtaining permission for follow-up.

Questions such as “Would you use this app?” are weak predictors because respondents are often polite and future intentions are uncertain. Questions about actual past behavior are generally more informative: “What did you do the last time this happened?”

Observation and contextual inquiry

Observation can reveal workarounds that participants forget or consider too ordinary to mention. A researcher should document actions, sequence, tools, interruptions, environmental constraints, and interactions—not merely impressions. Observation must respect privacy and organizational permissions.

Surveys

Surveys are useful for measuring the prevalence of patterns identified qualitatively. They are less useful when the researcher does not yet understand the vocabulary or response categories of the population. A survey should have a clear target population, neutral wording, mutually understandable response options, and a stated purpose.

A sample proportion is calculated as:

where x is the number of respondents selecting a response and n is the total number of valid respondents. The result describes the sample, not automatically the whole population. Generalization requires an appropriate sampling design and attention to sampling error, nonresponse, measurement bias, and population coverage.

A larger convenience sample does not automatically solve bias. One thousand responses from a self-selected online group may be less informative about a target population than 100 carefully sampled respondents.

Experiments and behavioral evidence

Behavioral evidence is often stronger than stated preference. Examples include a landing-page test, a paid pilot, a pre-order, a time-limited manual service, or an A/B test. Experiments should change one important variable where practical and define success criteria in advance. A click does not necessarily demonstrate willingness to pay, and a pre-order may be affected by incentives or unusual urgency.

5. Ethical Research Practice

Entrepreneurial research involves duties to participants. Ethical practice is not only a compliance requirement; it improves trust and data quality.

Researchers should:

• obtain informed consent;
• explain the purpose and foreseeable use of data;
• make participation voluntary and allow withdrawal where feasible;
• collect only data necessary for the research question;
• protect confidentiality and securely store records;
• avoid deceptive claims, coercion, and unreasonable incentives;
• disclose relevant conflicts of interest;
• use additional safeguards with children and vulnerable groups; and
• report findings honestly, including contradictory evidence.

Personal data should be separated from research notes when possible, access should be limited, and retention periods should be defined. Applicable privacy law varies by jurisdiction, but principles such as purpose limitation, data minimization, transparency, and security are broadly relevant. Sensitive data should not be collected merely because technology makes collection easy.

AI tools can assist with transcription, coding, clustering, and question generation, but researchers remain responsible for accuracy, confidentiality, bias, and consent. Sensitive interview recordings should not be uploaded to an external AI service unless authorized and appropriately protected. AI-generated themes must be checked against original evidence because automated systems can misinterpret context or reproduce stereotypes.

6. Secondary Research

Secondary research uses existing information, including government statistics, academic articles, trade publications, company filings, industry reports, regulatory documents, search data, and competitor materials.

A source should be evaluated for:

• authority: who produced it and what expertise or incentive do they have?
• method: how were data collected and defined?
• currency: how recent and time-sensitive is the information?
• coverage: does the population or geography match the venture?
• comparability: are categories and units consistent?
• bias: what claims or commercial interests may shape presentation?

Search volume is not the same as market demand, and a competitor’s promotional claims are not neutral market evidence. Secondary sources should be triangulated, meaning that important claims are compared across independent sources and, where possible, tested through primary research.

7. Market Sizing: TAM, SAM, and SOM

Market sizing provides an estimate of commercial opportunity, not a guarantee of revenue.

• Total Addressable Market (TAM): demand if the venture could serve the entire relevant market under defined conditions.
• Serviceable Available Market (SAM): the portion of TAM that fits the venture’s geography, product scope, channel, and capabilities.
• Serviceable Obtainable Market (SOM): the realistically attainable portion of SAM over a specified period, considering competition and execution constraints.

Two common approaches are the top-down approach and the bottom-up approach. Top-down estimates begin with an industry total and apply defensible filters. Bottom-up estimates build from customers, usage, and price, usually making assumptions more visible.

For a recurring-revenue offer, a simple annual revenue estimate is:

where N is the number of potential customers, f is average purchases per customer per year, and P is revenue per purchase. Units must be consistent: customers × purchases/customer/year × currency/purchase = currency/year.

If customers pay a monthly subscription, a simplified annual revenue estimate is:

where m is the number of paid months in the period and P_m is monthly price. This estimate assumes the number of customers, price, and retention pattern are represented adequately; it does not automatically account for churn, discounts, taxes, or variable usage.

Worked example

Suppose a venture targets 40,000 independent clinics. Each clinic could purchase an average of 10 software seats at $30 per seat per month. The theoretical annual TAM is:

40,000 × 10 × $30 × 12 = $144,000,000 per year.

If the venture initially serves only clinics in two regions representing 15% of the target base, the simplified SAM is:

$144,000,000 × 0.15 = $21,600,000 per year.

If the team believes it can acquire 2% of those regional clinics within an initial period, the corresponding revenue opportunity is:

$21,600,000 × 0.02 = $432,000 per year.

These figures are illustrative. They should be challenged with evidence about actual clinic counts, seat usage, willingness to pay, sales capacity, implementation constraints, and competing products. Multiplying large population figures by optimistic prices can produce impressive but strategically useless TAM estimates.

8. Competitor and Alternative Analysis

Competition includes more than businesses with similar branding. It includes direct competitors, indirect competitors, substitute products, internal processes, postponement, and doing nothing.

A competitor analysis should examine:

• target customer and use case;
• core job addressed;
• pricing and revenue model;
• distribution and sales channel;
• major benefits and limitations;
• switching costs and lock-in;
• evidence of traction or customer trust; and
• strategic vulnerabilities.

A comparison matrix is more useful when based on customer decision criteria rather than arbitrary feature counts. The question is not “Which product has more features?” but “Which alternative performs best for the job under the customer’s constraints?”

Competitor research should remain lawful and ethical. Public information, customer interviews, product trials, and published documents are legitimate sources. Misrepresentation, unauthorized access, theft of confidential information, and misuse of personal data are not.

9. Personas and Customer Insights

A persona is a research-based representation of a meaningful customer pattern. It is not a fictional character created from stereotypes. A useful persona includes context, goals, behaviors, triggers, barriers, decision criteria, and representative evidence.

A persona should distinguish:

• what customers say;
• what they do;
• what they believe or fear; and
• what constraints shape their choices.

Demographic detail should be included only when it affects the problem or decision. A persona named “Jordan, the time-pressed clinic manager” is useful if supported by evidence about scheduling, staffing, purchasing authority, and workflow. It is not useful merely because it has an age and a stock photograph.

From data to insight

A customer insight is an evidence-based interpretation that explains a meaningful pattern and suggests an action. The following distinction helps:

• Observation: six managers used spreadsheets to coordinate appointments.
• Interpretation: spreadsheets may be serving as a flexible coordination system, not merely a record-keeping tool.
• Insight: a replacement must preserve rapid, informal coordination or managers may resist switching.
• Implication: test a product that integrates with existing spreadsheets before requiring a complete workflow change.

Researchers should maintain an evidence log linking each major insight to interview excerpts, observations, survey results, or behavioral data. This prevents attractive narratives from becoming detached from evidence.

Figure: Customer discovery is an iterative process in which evidence revises assumptions and informs decisions.

10. An Integrated Research Process

A practical discovery process can be organized into seven stages:

1. Map assumptions. List beliefs about the customer, job, channel, price, and competition.
2. Rank uncertainty. Prioritize assumptions that are both important and poorly supported.
3. Define the research question. Specify what evidence is needed and what decision it will inform.
4. Select the method. Use interviews for depth, observation for behavior in context, surveys for prevalence, and experiments for behavioral response.
5. Recruit relevant participants. Include users, buyers, influencers, and non-adopters where appropriate.
6. Analyze systematically. Code recurring behaviors, problems, workarounds, and barriers; compare patterns across segments.
7. Decide and document. Record what was learned, what remains uncertain, and what action follows.

Research is iterative. A first set of interviews may reveal that the assumed customer is not the economic buyer. A survey may show that a problem is common but not urgent. A pilot may show that customers value the outcome but reject the proposed workflow. Each finding should update the venture’s model rather than be treated as a verdict on the founder’s identity or creativity.

Illustrative Case: FreshRoute’s Delivery Coordination Service

Background

FreshRoute is a fictional early-stage venture proposing a coordination service for small food retailers receiving deliveries from multiple suppliers. The founders initially believed that retailers needed route-optimization software.

Problem

The team had limited evidence. Its website visitors liked the concept, but no retailers had paid. The founders needed to determine whether route optimization was the core job and which segment experienced the strongest need.

Analysis and approach

The team mapped three assumptions: retailers had frequent delivery delays, managers were willing to purchase specialized software, and route optimization was the desired solution. It conducted 14 semi-structured interviews with independent grocers, specialty food shops, and small restaurant groups. It also observed receiving activities at four sites and reviewed publicly available logistics data.

The interviews showed that managers rarely described route optimization as their primary concern. Their repeated concern was uncertainty: they did not know when deliveries would arrive, which made staff scheduling and cold-storage preparation difficult. Existing alternatives included telephone calls, messaging applications, paper notes, and spreadsheets. The problem was most urgent for specialty grocers handling perishable products, but many owners lacked time to implement complex software.

The team then ran a manual pilot with five specialty grocers. Suppliers sent delivery updates to a FreshRoute operator, who consolidated the information into a simple dashboard and text alerts. The pilot did not test scalable technology; it tested whether timely information created value.

Outcome

Four of the five grocers used the alerts repeatedly during the pilot. Three agreed to pay for a monthly service if supplier participation could be improved. The team revised its job statement from “optimize delivery routes” to “help managers prepare for uncertain deliveries.” It postponed route optimization and tested supplier integration, alert accuracy, and a low-complexity onboarding process.

Learning points

• The original feature was not the same as the customer’s job.
• Observation exposed operational consequences that interviews alone might have missed.
• A manual pilot tested value before expensive software development.
• The strongest initial segment was defined by context and urgency, not simply business size.
• Willingness to pay became more credible when connected to repeated behavior and a concrete offer.

Challenges and Limitations

Research can mislead when participants are poorly selected, questions are leading, findings are generalized beyond the sample, or founders ignore inconvenient evidence. Early adopters may not represent mainstream customers. Stated intentions may differ from later behavior. Market-size estimates may double-count customers or confuse spending in adjacent categories with demand for the proposed offer. Competitor information may be incomplete or outdated.

Qualitative research produces depth rather than automatic prevalence estimates. Quantitative research can estimate patterns but may miss context and causal mechanisms. Neither method is inherently superior; the appropriate method depends on the decision and uncertainty.

Triangulation improves confidence but does not guarantee truth. Independent sources may share the same bias, and apparent agreement may reflect a common data error. Entrepreneurs should therefore communicate uncertainty explicitly, distinguish evidence from assumptions, and update conclusions when new information appears.

Professional and Managerial Implications

Customer discovery is an employability skill across entrepreneurship, product management, marketing, consulting, innovation, and nonprofit development. Professionals who can formulate neutral questions, analyze evidence, estimate opportunity, and communicate uncertainty contribute more effectively to cross-functional decisions.

Managers should create conditions in which negative findings are safe to report. If teams are rewarded only for confirming a launch decision, research becomes ceremonial. Useful governance includes research protocols, privacy reviews, decision logs, source documentation, and clear ownership for acting on findings.

Chapter Summary

Customer discovery and market research reduce entrepreneurial uncertainty by converting assumptions into questions and evidence. Effective segmentation identifies groups with sufficiently similar problems, contexts, and buying conditions. Jobs-to-be-done analysis focuses on the progress customers seek, including functional, emotional, and social dimensions, rather than prematurely defining a feature solution.

Primary research methods include interviews, observation, surveys, experiments, and pilots. Their value depends on appropriate participant selection, neutral questioning, careful interpretation, and recognition of bias. Secondary research extends discovery to industry structure, competitors, regulation, and market conditions; sources should be assessed for authority, method, currency, coverage, comparability, and bias.

TAM, SAM, and SOM provide increasingly focused views of opportunity, but market sizing is only as credible as its assumptions. Competitor analysis must include substitutes and inaction. Personas should represent evidence-based patterns, while insights should connect observations to implications and decisions. Ethical research requires consent, privacy protection, data minimization, honest reporting, and responsible use of AI. The central discipline is iterative learning: research should lead to a clearer decision to persist, adapt, or stop.

Key Terms

Customer discovery: A structured process for learning about customers, problems, behaviors, and adoption conditions.

Market research: The collection and analysis of information about customers, markets, competitors, and external conditions.

Customer segment: A group sharing relevant needs, behaviors, contexts, or buying conditions.

Jobs-to-be-done: The progress or outcome a customer seeks in a particular situation.

Functional job: A practical task or outcome a customer wants to accomplish.

Emotional job: The feeling or psychological state a customer seeks.

Social job: The identity or social impression a customer seeks to create.

Beachhead segment: A narrowly defined initial segment selected for focused entry and learning.

Primary research: New evidence collected directly from participants or observed settings.

Secondary research: Existing information collected by other researchers or organizations.

Triangulation: Comparing multiple sources or methods to strengthen interpretation.

Convenience sample: Participants selected because they are easy to access rather than representative.

TAM: Total Addressable Market under defined theoretical conditions.

SAM: Serviceable Available Market that fits the venture’s scope and constraints.

SOM: Serviceable Obtainable Market that the venture can realistically capture.

Competitor: A direct or indirect alternative addressing the same or related customer job.

Persona: A research-based representation of a meaningful customer pattern.

Customer insight: An evidence-based interpretation that explains a pattern and suggests action.

Informed consent: Voluntary agreement based on adequate information about research participation.

Confirmation bias: The tendency to favor evidence supporting existing beliefs.

Review Questions

Short Answer Questions

1. What is the difference between customer discovery and market research?
2. Why are questions about recent behavior generally more useful than questions about future intentions?
3. Distinguish TAM, SAM, and SOM.
4. Name three ethical responsibilities in primary research.
5. What is a job-to-be-done, and how does it differ from a product feature?

Long Answer Questions

1. Explain how a venture can move from assumptions to testable research questions. Illustrate your answer with a new business example.
2. Compare interviews, observation, surveys, and experiments in terms of their appropriate uses and limitations.
3. Explain how customer segmentation and jobs-to-be-done analysis can be combined to identify a beachhead segment.
4. Discuss why competitor analysis must include substitutes, internal workarounds, and doing nothing.
5. Evaluate the ethical and governance issues involved in using AI to analyze customer interviews.

Analytical / Application Questions

1. A student venture claims that all university students want a financial-planning application. Design a discovery study that tests the segment, job, current alternatives, and willingness to pay without leading participants.
2. Using the market-sizing equations in this chapter, construct a bottom-up TAM, SAM, and SOM estimate for a hypothetical subscription venture. State and defend every assumption.
3. Interviewees say they want a proposed service, but none has used the sign-up page or accepted a paid pilot. Analyze at least three possible explanations and recommend the next research action.

Multiple Choice Questions with Answers

Question 1

Which question is most appropriate for an early customer-discovery interview?

A. Would you definitely buy our proposed product?

B. What did you do the last time this problem occurred?

C. Do you agree that our solution is innovative?

D. How much would you pay for every feature?

Correct Answer: B. What did you do the last time this problem occurred?

Question 2

Which statement best describes a job-to-be-done?

A. A list of software features

B. A demographic profile of a customer

C. The progress a customer seeks in a particular situation

D. The legal structure of a venture

Correct Answer: C. The progress a customer seeks in a particular situation

Question 3

Which of the following is an example of a behavioral segmentation variable?

A. Age

B. Purchase frequency

C. Region

D. Household size

Correct Answer: B. Purchase frequency

Question 4

A market estimate uses 5,000 customers, 4 purchases per year, and a price of $25 per purchase. What is the estimated annual revenue opportunity?

A. $20,000

B. $50,000

C. $125,000

D. $500,000

Correct Answer: D. $500,000

Question 5

Which market-sizing category represents the realistically attainable portion of the serviceable market?

A. TAM

B. SAM

C. SOM

D. GDP

Correct Answer: C. SOM

Question 6

Why is a large convenience sample potentially misleading?

A. Convenience samples always contain too few participants

B. Sample size eliminates all forms of bias

C. Participants may not represent the target population

D. Convenience samples cannot contain numerical data

Correct Answer: C. Participants may not represent the target population

Question 7

Which practice best reflects informed consent?

A. Concealing the purpose of research to obtain natural responses

B. Explaining participation, data use, and voluntary withdrawal before participation

C. Collecting all available personal information for future use

D. Requiring participation as a condition of receiving a normal service

Correct Answer: B. Explaining participation, data use, and voluntary withdrawal before participation

Question 8

Which statement is an observation rather than an interpretation?

A. Managers resist all new technology

B. Spreadsheets are the best coordination system

C. Six managers used spreadsheets to coordinate appointments

D. Managers will purchase a replacement system

Correct Answer: C. Six managers used spreadsheets to coordinate appointments

Question 9

A competitor analysis should include which of the following?

A. Only firms with identical product features

B. Only competitors with the lowest prices

C. Direct competitors, substitutes, workarounds, and inaction

D. Only companies operating in another country

Correct Answer: C. Direct competitors, substitutes, workarounds, and inaction

Question 10

What is the main purpose of triangulation?

A. To guarantee that research findings are correct

B. To compare multiple sources or methods when interpreting evidence

C. To replace primary research with advertising data

D. To increase the number of features in a product

Correct Answer: B. To compare multiple sources or methods when interpreting evidence