Lean Startup, Prototyping and Validation

Introduction

Entrepreneurial ventures operate under conditions of uncertainty. Founders may not know whether a customer problem is important, whether customers will adopt a proposed solution, whether the venture can deliver it reliably, or whether revenue will exceed the cost of acquisition and operation. Traditional business planning often assumes that these questions can be answered in advance through market research and detailed forecasting. Lean startup practice takes a different position: many important assumptions can be tested more effectively through structured, low-cost experiments with real or representative customers.

Lean startup is not simply a method for building products quickly. It is a disciplined approach to learning. It combines design thinking, hypothesis development, rapid prototyping, minimum viable products (MVPs), customer evidence and iterative decision-making. Its purpose is to reduce the most important uncertainties before the venture commits substantial time, money and organizational resources.

The central logic is the build–measure–learn cycle. A team builds an experiment or product representation, measures customer and operational evidence, learns whether its assumptions are supported, and then decides whether to persevere, modify the model or abandon a particular direction. Effective validation therefore requires more than collecting opinions. It requires explicit hypotheses, observable evidence, appropriate metrics and predetermined decision rules.

Learning Objectives

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

1. Explain the principles and assumptions of lean startup methodology.
2. Distinguish desirability, feasibility and viability questions.
3. Convert business-model assumptions into testable hypotheses.
4. Select appropriate prototypes and MVPs for different learning objectives.
5. Design staged experiments and interpret evidence using the build–measure–learn cycle.
6. Apply quantitative validation metrics while recognizing their limitations.
7. Evaluate ethical, legal and practical risks in entrepreneurial experimentation.
8. Make evidence-based decisions about persevering, pivoting or stopping.

1. Foundations of Lean Entrepreneurship

1.1 Uncertainty as the central management problem

A business model contains assumptions about customers, problems, value propositions, channels, relationships, resources, partners, activities, costs and revenues. Some assumptions are relatively certain; others are conjectures. A founder who treats conjectures as facts may invest heavily in a product that customers do not value or in a delivery system that cannot scale economically.

Lean entrepreneurship separates execution risk from search risk. Execution risk concerns performing a known activity well, such as manufacturing a proven product at a target cost. Search risk concerns discovering what should be built, for whom, and under what economic conditions. Startups typically face high search risk, so learning speed and learning quality are strategic capabilities.

The lean approach is associated with several complementary traditions:

• Design thinking emphasizes understanding users, framing problems, generating alternatives, prototyping and testing.
• Effectuation views entrepreneurs as acting with available means and adapting to contingencies rather than relying entirely on prediction.
• The scientific method emphasizes explicit propositions, observable evidence and revision in response to results.
• Agile development emphasizes incremental delivery, feedback and adaptation in product development.
• Lean production emphasizes reducing waste, improving flow and creating value from the customer’s perspective.

Lean startup integrates these ideas for situations in which both the problem and the solution may be uncertain.

1.2 Core principles

Five principles are especially important.

Entrepreneurs are testing a business model, not merely constructing a product. A technically successful product can still fail because the target customer is too narrow, acquisition is too expensive, regulation prevents deployment, or revenue does not support the cost structure.

Learning must be validated. A team gains validated learning when evidence demonstrates whether a meaningful assumption is supported in a defined context. General enthusiasm, social-media attention or internal belief is not equivalent to validated learning.

Small experiments reduce irreversible commitment. Early tests should be designed to obtain important information with limited expenditure. This is not an argument for poor quality; it is an argument for appropriate fidelity.

Progress is measured by uncertainty reduction. In the earliest stages, revenue may be a poor indicator of progress if the product is not yet available. Evidence that identifies the strongest customer segment, rejects an attractive but false assumption, or demonstrates a workable process may be more valuable.

Iteration must be disciplined. Repeated changes without a clear hypothesis create random motion. Each iteration should state what is being tested, what evidence will be collected, and what decision will follow.

2. Design Thinking and the Three Validation Questions

Design thinking is commonly represented as a non-linear sequence of empathize, define, ideate, prototype and test. In entrepreneurship, these activities should be understood as mutually reinforcing rather than strictly sequential. Testing can reveal that the original problem was framed incorrectly; prototyping can reveal unrecognized user needs; and customer research can generate new solution concepts.

A business model should be assessed through three related dimensions:

These dimensions interact. A desirable product that cannot be delivered reliably is not feasible. A feasible product that customers will not pay for is not viable. A financially attractive model that harms users or violates regulation is unacceptable, regardless of its apparent profitability.

2.1 Problem–solution fit and product–market fit

Problem–solution fit concerns whether a proposed solution addresses a significant customer problem. It is usually explored before extensive investment in a complete product. Product–market fit is a stronger and later-stage condition in which a product demonstrates sustained demand within a defined market and can grow without proportionate increases in effort.

These concepts should not be treated as binary events. They represent increasing confidence based on evidence. A founder may have strong evidence of problem significance but weak evidence of willingness to pay. Another venture may have early adoption but poor retention. Validation is therefore multidimensional and stage-dependent.

3. Hypothesis Development

3.1 From assumptions to hypotheses

An assumption is an unverified belief. A hypothesis converts that belief into a proposition that can be tested.

A useful hypothesis specifies:

1. the target customer or context;
2. the problem or behavior;
3. the proposed intervention or value proposition;
4. the expected measurable response;
5. the time period or conditions;
6. the evidence threshold for action.

For example, “Students need affordable healthy meals” is too broad. A more testable proposition is: “Among first-year students living on campus, at least 30% of visitors to a mobile pre-order page will submit a meal reservation for a lunch priced at £6.50 during a two-week test.” This statement identifies a segment, behavior, offer, price, period and threshold.

3.2 Assumption prioritization

Not every assumption deserves immediate testing. Teams should prioritize assumptions according to both uncertainty and consequence. A highly uncertain assumption with little effect on the model can wait; a highly uncertain assumption that could invalidate the entire model should be tested early.

A practical prioritization matrix uses two dimensions:

• Importance: How damaging would it be if the assumption were false?
• Confidence: How strong is the existing evidence?

The highest-priority assumptions are important and weakly supported. Typical examples include whether customers recognize the problem, whether they will change current behavior, whether a critical input can be obtained at the required cost, and whether a regulator permits the proposed activity.

3.3 Falsifiability and evidence quality

A hypothesis must be capable of being contradicted. “Customers will love our product” is not falsifiable because “love” is undefined. A measurable version might specify a repeat-use rate, pre-order rate or conversion rate.

Evidence quality depends on several factors:

• Behavior versus intention: Actual payment, use or referral is generally stronger evidence than stated interest.
• Representativeness: Evidence from the target segment is more useful than evidence from friends or unrelated respondents.
• Sample size and consistency: A single observation can generate insight but rarely establishes a dependable pattern.
• Context: Results from a highly subsidized test may not predict behavior at normal prices.
• Measurement validity: The metric must reflect the underlying question rather than an easy-to-count substitute.

4. Prototyping as a Learning Instrument

A prototype is a representation of a product, service or business process used to explore assumptions. It may be physical, digital, visual, experiential or operational. The appropriate prototype is the simplest one that can produce credible evidence for the current question.

4.1 Types of prototypes

Sketches and storyboards are useful for exploring user journeys, service interactions and alternative concepts. They are inexpensive but provide limited evidence about technical performance.

Wireframes and clickable interfaces test navigation, information structure and task comprehension before software is built.

Role-playing and service simulations allow teams to examine front-stage and back-stage activities, such as booking, delivery, customer support and exception handling.

Landing pages and smoke tests examine whether a clearly defined audience responds to a value proposition. They can measure sign-ups or requests for more information, but they do not by themselves prove sustained usage or profitability.

Concierge MVPs deliver a service manually to a small number of customers. They are useful when the main uncertainty concerns customer value rather than automation or scale.

Wizard-of-Oz prototypes present an apparently automated interface while humans perform the underlying operation. They can test customer interaction before expensive engineering, provided that participants are not misled about material matters.

Technical prototypes test performance, integration, reliability, safety or manufacturability. A technical demonstration may establish feasibility without proving desirability.

4.2 Minimum viable product

An MVP is the smallest product or service configuration that enables a team to test a specific value proposition with real users and obtain meaningful evidence. “Minimum” refers to scope, not carelessness; “viable” means sufficiently functional to deliver the learning objective and an acceptable user experience.

An MVP should not be defined as a universally small product. Its required scope depends on the hypothesis. If the uncertainty concerns taste, a manually prepared sample may be appropriate. If it concerns data security, a landing page is inadequate and a controlled technical test is required.

Common errors include:

• building many features before testing the core problem;
• confusing a prototype with a complete business model;
• using an MVP to expose users to unsafe or unreliable conditions;
• measuring clicks while ignoring retention and payment;
• interpreting a temporary promotion as evidence of normal demand.

5. The Build–Measure–Learn Cycle

The cycle begins with a business-model assumption and ends with a decision. It should be treated as a learning loop rather than a production schedule.

5.1 Build

The team builds the smallest credible experiment. It defines the target participant, interaction, offer, duration, resources and safeguards. A build may be a product, a service script, a web page, a pricing conversation or a technical test.

5.2 Measure

Measurement involves collecting evidence linked to the hypothesis. Useful metrics include:

• Conversion rate: the proportion of eligible prospects who complete a specified action.
• Activation rate: the proportion of new users who reach a meaningful first-value event.
• Retention rate: the proportion of users who return or remain active during a defined period.
• Churn rate: the proportion of customers or users lost during a defined period.
• Customer acquisition cost (CAC): acquisition expenditure divided by the number of new customers attributed to that expenditure.
• Contribution margin: revenue per unit minus variable cost per unit.

For example, if 240 eligible visitors produce 36 completed sign-ups, then:

This calculation assumes that the denominator includes only eligible visitors and that each sign-up is counted once. The result describes this experiment; it does not automatically generalize to all future traffic.

If acquisition spending is £900 and it produces 60 new paying customers, then:

If average revenue per order is £25 and variable cost per order is £10, contribution margin per order is:

A simplified comparison suggests that one order does not recover a £15 CAC before fixed costs. However, customer lifetime, repeat purchases, refunds, support costs and attribution uncertainty must be considered before making a viability decision.

5.3 Learn

Learning requires interpretation, not merely reporting. The team should compare results with the predetermined threshold, investigate unexpected behavior and assess alternative explanations. If a test fails, the conclusion may be that the problem is weak, the segment is wrong, the offer is unclear, the price is unsuitable, the channel is ineffective or the experiment was poorly designed.

5.4 Persevere, pivot or stop

A persevere decision means continuing with the current central model while improving execution. A pivot is a substantive, evidence-based change in one or more elements of the model, such as customer segment, problem, channel, revenue model or technology. A stop decision ends a line of inquiry when evidence indicates that further investment is not justified.

A pivot should preserve learning while changing a critical assumption. It is not merely changing a color, slogan or minor feature.

6. Staged Experiments and Decision Architecture

A staged validation program reduces the risk of testing expensive assumptions with expensive methods. A useful sequence is:

1. Discovery stage: interviews, observation and problem research establish context and language.
2. Concept stage: sketches, storyboards and value-proposition tests explore solution relevance.
3. Behavior stage: landing pages, pre-orders, concierge delivery or trials test action.
4. Feasibility stage: technical, operational, supplier and compliance tests examine delivery capability.
5. Viability stage: pricing, retention, unit economics and repeatable acquisition are examined.
6. Scale stage: processes, capacity, governance and financing are tested under increasing volume.

The stages need not be strictly linear. A safety-critical product may require feasibility testing before user trials. A regulated financial service may require legal analysis before collecting customer data.

A strong experiment brief includes:

Predetermined thresholds reduce confirmation bias. Teams should also record adverse evidence and unexpected behavior rather than selecting only favorable observations.

7. Quantitative Reasoning in Validation

Quantitative metrics support judgment but do not replace it. A rate without a clear denominator, time frame and population can be misleading. Teams should distinguish descriptive results from causal conclusions.

For a simple experiment with n independent observations and x observed successes, the sample proportion is:

Here, p̂ is the observed proportion, x is the number of successes and n is the number of eligible observations. This formula describes the sample; it does not guarantee that the true population proportion equals p̂. Small or biased samples may produce unstable estimates.

When comparing two versions, such as two prices or messages, the team may calculate the difference in observed conversion rates:

where p̂_A and p̂_B are the observed rates for versions A and B. A difference may result from the intervention, random variation, different audiences, timing or other confounding factors. Random assignment, consistent conditions and adequate sample size improve interpretation.

For financial viability, a simplified break-even quantity is:

This assumes one product or service, constant price and variable cost per unit, and a stable relationship between volume and costs. If fixed costs are £30,000 and contribution margin is £15 per order, break-even volume is:

The calculation excludes taxes, financing costs, capacity constraints and changes in price or cost. It is a planning approximation, not a prediction.

8. Applied Case: FreshRoute Campus Meals

The following is an illustrative case.

Background

FreshRoute is a proposed venture offering pre-ordered, affordable meals to university students. The founding team believes that students want healthier food but face long queues, uncertain availability and limited budgets. The initial business model assumes that students will order through a mobile interface, collect meals at a campus location and pay £6.50 per meal.

Problem and assumptions

The team identifies four high-risk assumptions:

1. Students consider the problem important enough to change their current behavior.
2. They will pay £6.50 rather than use existing food outlets.
3. Meals can be prepared and distributed within the lunch period.
4. Sufficient gross contribution remains after ingredients, packaging, payment fees and labor.

The team does not begin by building a full application. It first conducts observational research at lunch locations and interviews students about actual purchasing behavior. It discovers that “healthy” is less important than speed, predictable price and vegetarian availability for much of the target segment.

Experiment 1: Problem and segment discovery

The team observes 80 lunch purchases and interviews 24 students. It records queues, waiting time, price, dietary requirements and whether students abandon a purchase. Interviews are used to understand context, while observations reduce reliance on stated intentions.

The evidence suggests that commuters and students with classes immediately after lunch experience the greatest pain. The team narrows its initial segment to students with a 30-minute timetable gap.

Experiment 2: Behavioral demand test

A simple mobile webpage presents three meals, collection times and a £6.50 price. Rather than accepting vague expressions of interest, the team requests a refundable £1 reservation deposit for a one-week pilot. Of 150 qualified visitors, 39 submit deposits.

The observed reservation rate is:

This result supports initial desirability, but it does not prove repeat demand. The deposit may also be influenced by novelty or the refund option.

Experiment 3: Concierge MVP

For one week, the founders manually prepare and distribute 25 meals per day. They use a spreadsheet instead of building an application. They measure on-time collection, cancellations, repeat orders, complaints and preparation time.

The pilot reveals that demand is strongest on days with consecutive classes, while late collection creates food-waste and staffing problems. The manual service also shows that customers value a guaranteed collection window more than menu variety.

Experiment 4: Viability analysis

The average price is £6.50. Ingredients and packaging cost £2.70 per meal, payment fees average £0.20, and variable labor is £1.10. Contribution margin is therefore:

If weekly fixed operating costs are £1,000, the simplified weekly break-even volume is:

The pilot volume of 125 meals per week is far below this level. The team therefore cannot justify immediate scaling. It investigates whether a higher-volume collection point, employer subsidy, a £7.25 premium option or reduced preparation labor could improve the model.

Outcome

FreshRoute does not abandon the concept, but it pivots from a broad “healthy student meals” proposition to a narrower “guaranteed 10-minute collection for students with compressed schedules.” It delays custom software, tests a campus partnership and limits the menu to items that can be prepared consistently. The evidence has not established a complete business, but it has reduced uncertainty about the segment, value proposition, operations and economics.

Learning points

The case illustrates five principles. First, customer language may differ from the founder’s initial framing. Second, a deposit provides stronger evidence than a survey response. Third, a concierge MVP can test value before automation. Fourth, desirability does not guarantee viability. Fifth, numerical thresholds help prevent premature scaling.

9. Challenges, Ethics and Governance

Lean experiments can create risks when teams value speed more than responsibility. Participants should understand what they are being asked to do, what data are collected and what risks exist. Informed consent is particularly important when experiments involve health, finance, employment, vulnerable populations or sensitive personal information.

Data protection obligations may include lawful collection, purpose limitation, data minimization, security, retention controls and rights of access or deletion. Teams should not collect personal data merely because future use seems possible. A/B testing must also avoid discriminatory outcomes, deceptive interfaces and unfairly different treatment of vulnerable users.

A prototype that is acceptable for a low-risk consumer service may be inappropriate in medical, transport, childcare or financial contexts. Safety, accessibility and regulatory review must be matched to the consequences of failure. “It is only an MVP” is not a defense for preventable harm.

Common methodological limitations include selection bias, novelty effects, survivorship bias, small samples, seasonality, demand created by discounts and the Hawthorne effect, in which participants alter behavior because they know they are observed. Founders should document these limitations and avoid presenting early results as universal truths.

10. Technology, AI and Contemporary Practice

Cloud services, no-code tools, analytics platforms, digital payments and generative AI can reduce the cost and time of prototyping. AI can help summarize interview transcripts, generate alternative concepts, simulate user conversations or identify patterns in customer feedback. However, AI-generated interpretations may reproduce bias, hallucinate evidence or obscure the difference between a customer’s actual statement and an inferred theme. Human review, secure data handling and traceable research records remain necessary.

Digital experimentation also creates risks of surveillance and manipulation. Tracking should be proportionate to the learning objective. Teams should distinguish personalization that helps users from dark patterns that pressure them into actions they did not intend. Automated decisions should be explainable and monitored for unequal effects.

Contemporary validation increasingly includes sustainability and resilience. A model should be tested not only for immediate customer demand but also for material use, energy consumption, labor conditions, supply-chain dependence, repairability and end-of-life consequences. A venture that appears viable only because it externalizes environmental or social costs has an incomplete business model.

11. Best Practices for Entrepreneurial Validation

1. Start with the highest-risk assumption, not the easiest feature to build.
2. Define the customer and context precisely.
3. Use behavior, payment and repeat use as stronger evidence than enthusiasm alone.
4. Match prototype fidelity to the question being tested.
5. Set metrics and decision thresholds before observing results.
6. Separate discovery research from promotional selling.
7. Track negative evidence, dropouts and unintended consequences.
8. Test desirability, feasibility and viability separately and together.
9. Protect participant privacy and design for accessibility and safety.
10. Record experiment conditions so results can be interpreted and reproduced.
11. Use cohorts and time-based retention rather than relying only on cumulative totals.
12. Treat a pivot as a reasoned change in a business-model assumption, not as random experimentation.
13. Scale only after the critical process works at an economically credible level.
14. Communicate uncertainty honestly to co-founders, employees, investors and customers.

Professional and Employability Relevance

Validation capability is valuable in entrepreneurship, product management, consulting, innovation, marketing, operations and corporate venture development. Employers expect graduates to frame ambiguous problems, conduct customer research, design experiments, interpret data, communicate trade-offs and make decisions without overstating certainty. These skills complement technical competence: a product manager must know what to build, an analyst must know which metric matters, and a founder must know when evidence justifies further investment.

Chapter Summary

Lean startup is a disciplined approach to reducing entrepreneurial uncertainty through explicit assumptions, rapid but responsible experiments and validated learning. Design thinking contributes empathy, problem framing and prototyping; the scientific method contributes falsifiable hypotheses and evidence; agile practice contributes iterative delivery.

Business-model validation addresses desirability, feasibility and viability. Hypotheses should identify a target segment, expected behavior, conditions, metrics and decision thresholds. Prototypes and MVPs should be selected according to the uncertainty being tested, not according to a fixed definition of “small.” The build–measure–learn cycle transforms assumptions into experiments, evidence and decisions to persevere, pivot or stop.

Quantitative measures such as conversion, retention, CAC, contribution margin and break-even volume improve decision quality when their denominators, assumptions and limitations are clear. Early evidence is often incomplete and subject to bias, so ethical safeguards, privacy protection, regulatory compliance and transparent communication are essential. Effective validation does not eliminate uncertainty; it makes uncertainty visible and supports better allocation of entrepreneurial resources.

Key Terms

Assumption: An unverified belief about a customer, product, operation or business model.

Build–measure–learn: An iterative cycle in which a team creates an experiment, collects evidence and revises its understanding.

Concierge MVP: A manually delivered service used to test customer value before automation.

Conversion rate: The proportion of eligible prospects who complete a specified action.

Desirability: The extent to which customers want or value a proposed solution.

Feasibility: The extent to which a solution can be delivered technically, operationally and legally.

Hypothesis: A specific, measurable and potentially falsifiable proposition.

Lean startup: A method for developing ventures through rapid, evidence-based learning under uncertainty.

Minimum viable product: The smallest sufficiently functional offering that can test a defined business assumption.

Pivot: A substantive, evidence-based change to a business-model element.

Problem–solution fit: Evidence that a proposed solution addresses a significant customer problem.

Prototype: A representation used to explore or test a product, service or process.

Retention: The proportion of users who continue to use or purchase during a defined period.

Smoke test: A low-cost test of interest or intended action before the full offering exists.

Validated learning: Evidence-based understanding of whether a meaningful assumption is supported.

Viability: The capacity of a venture to create and capture sufficient economic value.

Wizard-of-Oz prototype: A seemingly automated experience whose underlying operations are performed manually.

Unit economics: The revenue, variable cost and contribution generated by a single customer or transaction.

Review Questions

Short Answer Questions

1. What is the difference between execution risk and search risk?
2. Define a minimum viable product and explain why it is not necessarily a low-quality product.
3. What are the three dimensions of desirability, feasibility and viability?
4. Why is a falsifiable hypothesis superior to a general statement of customer interest?
5. State two limitations of using conversion rate as evidence of product success.

Long Answer Questions

1. Explain the build–measure–learn cycle and discuss how it supports evidence-based entrepreneurial decision-making.
2. Compare sketches, concierge MVPs, landing pages and technical prototypes. Identify the type of uncertainty each is best suited to test.
3. Discuss how design thinking and lean startup complement one another in the development of a new venture.
4. Explain why a venture can demonstrate strong desirability but weak viability. Use suitable metrics in your answer.
5. Evaluate the ethical and regulatory responsibilities of a startup conducting digital experiments with customer data.

Analytical / Application Questions

1. A venture receives 48 paid reservations from 320 qualified visitors. Calculate the reservation rate and explain why the result alone cannot establish product–market fit.
2. A service charges £40 per transaction and has variable costs of £24. Monthly fixed costs are £12,800. Calculate the contribution margin per transaction and simplified break-even volume. State two assumptions underlying the calculation.
3. Design a staged experiment plan for a proposed peer-to-peer tutoring platform. Identify one desirability, one feasibility and one viability hypothesis, together with a prototype, metric and decision rule for each.

Multiple Choice Questions with Answers

Question 1

Which statement best describes validated learning?

A. An internal team belief that the product is attractive

B. Evidence-based understanding of whether a meaningful assumption is supported

C. A completed product with every planned feature

D. A large amount of website traffic without behavioral analysis

Correct Answer: B. Evidence-based understanding of whether a meaningful assumption is supported

Question 2

Which prototype is most appropriate for testing whether customers value a service before automation is built?

A. A concierge MVP

B. A final production system

C. A financial audit

D. A trademark application

Correct Answer: A. A concierge MVP

Question 3

A test receives 40 completed sign-ups from 200 eligible visitors. What is the conversion rate?

A. 5%

B. 10%

C. 20%

D. 40%

Correct Answer: C. 20%

Question 4

Which question primarily concerns feasibility?

A. Will customers pay for the proposed solution?

B. Can the venture deliver the solution reliably and legally?

C. Do customers recognize the problem?

D. Will customers recommend the product to friends?

Correct Answer: B. Can the venture deliver the solution reliably and legally?

Question 5

What makes a hypothesis falsifiable?

A. It uses enthusiastic language

B. It is supported by the founder’s intuition

C. It specifies observable evidence that could contradict it

D. It avoids measurable targets

Correct Answer: C. It specifies observable evidence that could contradict it

Question 6

A product has revenue of £30 per unit and variable cost of £18 per unit. What is its contribution margin per unit?

A. £12

B. £18

C. £30

D. £48

Correct Answer: A. £12

Question 7

Which result is generally stronger evidence of demand than a survey response saying that a customer is interested?

A. A founder’s personal opinion

B. A social-media impression

C. A completed payment or reservation

D. An unprompted internal forecast

Correct Answer: C. A completed payment or reservation

Question 8

What is a pivot?

A. A minor visual change with no change in assumptions

B. A substantive, evidence-based change to a business-model element

C. A decision to continue without measuring results

D. A method of calculating conversion rate

Correct Answer: B. A substantive, evidence-based change to a business-model element

Question 9

If fixed costs are £10,000 and contribution margin is £20 per unit, what is the simplified break-even quantity?

A. 200 units

B. 500 units

C. 2,000 units

D. 20,000 units

Correct Answer: C. 2,000 units

Question 10

Why should experiment thresholds be defined before results are observed?

A. To eliminate the need for customer research

B. To reduce confirmation bias and clarify decisions

C. To guarantee a positive outcome

D. To make every prototype fully automated

Correct Answer: B. To reduce confirmation bias and clarify decisions