Research Problems, Questions, and Decision Contexts

Introduction

Market research becomes valuable only when it informs a decision. Organisations rarely commission research because they merely want “more data.” They commission it because a product is underperforming, a market opportunity is uncertain, customers are switching, a new proposition must be designed, or a strategic choice carries significant risk. The central professional skill is therefore not simply collecting information; it is translating an ambiguous business situation into a disciplined research problem that can produce decision-relevant evidence.

This chapter explains how product managers and researchers move from a business challenge to a decision problem, research objectives, research questions, hypotheses, information requirements, scope, timelines, budgets, and a research brief. It also shows why poorly framed research can generate technically impressive but strategically useless findings.

Learning Objectives

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

1. Distinguish a business problem, decision problem, research problem, research question, objective, and hypothesis.
2. Diagnose the decision context, including stakeholders, uncertainty, consequences, constraints, and timing.
3. Convert managerial concerns into clear research objectives and answerable questions.
4. Develop hypotheses and information requirements without prematurely prescribing a method.
5. Define scope, population, geography, time horizon, deliverables, schedule, and budget.
6. Prepare and evaluate a professional research brief.
7. Recognise common framing errors, ethical risks, and the implications of AI-assisted research planning.

1. From Business Situation to Researchable Problem

A business situation is an observed condition requiring attention. Examples include declining activation, low repeat purchase, high customer-support contact, weak conversion, or uncertainty about entering a market. A situation is not automatically a research problem. Research is justified only when relevant information is unavailable, unreliable, or insufficient for a decision.

A useful chain is:

Business situation → decision problem → research problem → research questions → evidence → decision and action

1.1 Business situation

The business situation describes what is happening and why it matters. It should include evidence where available, rather than assumptions. “Sales are declining” is weaker than “Monthly paid conversion fell from 8.2% to 5.9% over the last two quarters, while traffic remained broadly stable.”

1.2 Decision problem

The decision problem states what management must decide. It is action-oriented and normally includes alternatives. For example:

Should the product team simplify onboarding, improve acquisition targeting, revise pricing, or maintain the current design?

A decision problem is not “understand customers.” It identifies a choice, the decision owner, and the consequences of acting or not acting.

1.3 Research problem

The research problem specifies the knowledge gap that prevents a sound decision. For example:

The organisation does not know whether low conversion is primarily caused by perceived complexity, weak value communication, trust concerns, poor traffic quality, or price sensitivity.

A good research problem is neither so broad that it becomes unmanageable nor so narrow that it excludes important explanations.

1.4 Research questions

Research questions translate the knowledge gap into answerable enquiries. They may ask about description, behaviour, causes, meanings, relationships, segments, or likely responses. Examples include:

• What tasks do new users attempt during their first session?
• At which stage do users abandon onboarding?
• How do users interpret the product’s value proposition?
• Which barriers distinguish activated from non-activated users?
• How would alternative onboarding concepts affect perceived effort and confidence?

A question should identify the population, phenomenon, context, and relevant time period where necessary.

2. The Decision Context

Research design depends on the context in which findings will be used. Two studies may investigate the same customer behaviour but require different designs because the decisions, risks, timelines, and stakeholders differ.

2.1 Decision owner and stakeholders

The decision owner has authority to act on the findings. Stakeholders may include product, marketing, sales, finance, engineering, legal, customer service, and external partners. Their interests may conflict. Marketing may want evidence for positioning, engineering may need implementation requirements, and finance may require a defensible revenue case.

The researcher should distinguish:

• Decision owner: accountable for the final choice.
• Users of findings: people who interpret or apply the results.
• Subjects or participants: people who provide data.
• Gatekeepers: people controlling access to participants, systems, or data.
• Affected parties: people who may experience consequences of the decision.

2.2 Decision alternatives

A study should be designed around plausible alternatives, including “do nothing,” delay, pilot, scale, or discontinue. If management has already selected a solution, research may be used merely to justify it. This creates confirmation bias.

A decision matrix can clarify the context:

2.3 Uncertainty and decision risk

Not every uncertainty deserves research. A useful test is whether resolving the uncertainty could materially change the decision. Research priority rises when uncertainty is high, the decision is difficult to reverse, the affected population is large, or the consequences of error are substantial.

A simplified expected-loss perspective is useful. Suppose an organisation chooses between options A and B under uncertain conditions. The expected loss of a decision can be represented conceptually as:

Here, E(L) is expected loss, s is a possible state of the world, p(s) is the estimated probability of that state, d is the chosen decision, and L(d,s) is the loss from taking decision d when state s occurs. This is a decision-analysis concept, not a claim that research can always produce precise probabilities. Its practical implication is that research is most valuable when it can reduce uncertainty that has meaningful consequences.

2.4 Exploratory, descriptive, diagnostic, and predictive purposes

Research purposes are often combined, but distinguishing them improves design.

• Exploratory research investigates poorly understood meanings, needs, motivations, and possible explanations. It is often qualitative and open-ended.
• Descriptive research estimates prevalence, frequency, profile, satisfaction, awareness, or behaviour in a defined population.
• Diagnostic or explanatory research examines why a pattern occurs and whether variables are associated.
• Predictive or evaluative research estimates likely responses to alternatives or assesses the performance of an intervention.

These labels do not determine method by themselves. A survey can be exploratory if it uses open-ended questions; interviews can be evaluative if participants assess a prototype. The purpose should be established before selecting the method.

3. Translating Objectives into Research Questions

3.1 Research objectives

A research objective states what the study will accomplish. Strong objectives use observable verbs such as identify, estimate, compare, assess, map, evaluate, explain, prioritise, or test. Weak objectives use vague verbs such as understand or explore without specifying the object of understanding.

A practical objective structure is:

To [verb] [phenomenon or variable] among [population] in [context] during [period], in order to inform [decision].

For example:

To identify the main barriers to first-week activation among newly registered users in the Indian small-business segment, in order to prioritise onboarding improvements for the next release.

A study may have one primary objective and several secondary objectives. The primary objective should map directly to the decision. Secondary objectives should not consume resources unless they have a credible use.

3.2 The hierarchy of questions

A useful hierarchy moves from broad framing to operational questions:

1. Context question: What is happening?
2. Behaviour question: What do people do?
3. Experience question: What do they experience or perceive?
4. Explanatory question: Why might the pattern occur?
5. Evaluation question: How do alternatives perform?
6. Decision question: What should the organisation do?

The final decision question is usually answered by combining research evidence with commercial, technical, financial, and regulatory information. Research should not pretend to make decisions that require managerial judgement.

3.3 Quality criteria for research questions

A good question is:

• Relevant: connected to the decision.
• Specific: clear about population, context, and phenomenon.
• Answerable: capable of being addressed with feasible evidence.
• Neutral: does not imply a preferred answer.
• Non-duplicative: adds distinct value.
• Prioritised: distinguishes essential from desirable information.
• Ethically acceptable: does not require unjustified intrusion or deception.

“Do customers love our new feature?” is leading and vague. “How does the feature affect perceived usefulness, effort, and willingness to continue using the product among current users?” is more precise, although it may still need a defined evaluation design.

4. Hypotheses and Conceptual Models

A hypothesis is a provisional, testable statement about a relationship, difference, or expected outcome. It is not a prediction inserted merely to make a study appear scientific. Qualitative research may use sensitising concepts or propositions rather than formal hypotheses.

Examples include:

• Users who perceive onboarding as easier are more likely to complete activation.
• Price sensitivity is higher among occasional users than among high-frequency users.
• Trust concerns are more common among first-time users than among existing customers.

A useful hypothesis identifies constructs and an expected relationship. Constructs are abstract concepts such as perceived value, trust, effort, involvement, or switching intention. They must be operationalised through observable indicators.

For a quantitative test, a null hypothesis may be stated as:

and a corresponding two-sided alternative as:

Here, μ₁ and μ₂ are population means for two defined groups. The notation is applicable when the outcome is measured on a scale for which comparison of means is meaningful and when the study design and statistical assumptions support the chosen test. A statistically detectable difference does not automatically establish causation or managerial importance.

4.1 Causal caution

Research questions often use causal language too early. “Does the new interface increase retention?” requires a design capable of identifying the effect of the interface while accounting for alternative explanations. An observational association may show that users exposed to the interface have higher retention, but those users may differ in motivation or prior experience.

Causal claims require attention to temporal order, plausible mechanisms, confounding variables, comparison conditions, and implementation fidelity. When randomised experiments are not possible, quasi-experimental or carefully controlled observational designs may provide useful evidence, but their limitations must be explicit.

5. Information Requirements and Operationalisation

Information requirements specify what must be known to answer the research questions and support the decision. They form the bridge between conceptual objectives and data collection.

5.1 Variables and measures

A variable is a characteristic that can take different values. A measure is the operational procedure used to represent it. “Trust” is a construct; agreement with carefully designed trust items, observed disclosure behaviour, or repeat usage may be possible indicators. No single measure should automatically be treated as the construct itself.

Measurement decisions should specify:

• Unit of analysis: person, account, household, transaction, session, or organisation.
• Population: the full set to which conclusions are intended to apply.
• Sample: the units actually studied.
• Time frame: when behaviour or attitudes are measured.
• Scale and coding: categories, ranks, counts, ratings, or continuous values.
• Inclusion and exclusion criteria.

5.2 Prioritising information

Information can be classified as:

• Must know: essential to the decision.
• Should know: useful if time and budget permit.
• Nice to know: interesting but not decision-critical.

This prevents scope expansion and protects analytical quality. A research brief should explain what the study will not answer. Explicit exclusions are a sign of discipline, not weakness.

6. Scope, Feasibility, Timeline, and Budget

6.1 Scope definition

Scope should cover population, geography, product or category, customer stage, channels, time period, methods, outputs, and decision boundary. For example, a study may cover active users of a business application in three cities who registered in the previous six months; it may exclude enterprise accounts and former users because separate research is planned.

Scope is also a governance mechanism. Without it, stakeholders may add questions after recruitment, change the target population, or expect causal conclusions from descriptive data.

6.2 Timeline planning

A research timeline normally includes:

1. Alignment and brief approval.
2. Secondary research and data audit.
3. Instrument or discussion-guide development.
4. Pilot and revision.
5. Recruitment or sample preparation.
6. Fieldwork or data extraction.
7. Quality checks and analysis.
8. Synthesis, reporting, and decision workshop.

Dependencies matter. Recruitment cannot begin until eligibility criteria and materials are approved; analysis cannot be finalised until data quality is assessed. A fast study is not necessarily an efficient study if compressed piloting produces invalid evidence.

6.3 Budget categories

A research budget may include internal labour, agency or panel fees, participant incentives, travel, software, translation, transcription, data engineering, accessibility support, legal review, and contingency. Costs should be linked to decisions and risks rather than treated as isolated line items.

A useful budget structure is:

The terms represent total budget and its listed components. This is an accounting identity for the specified categories, not a universal cost model. Categories must be defined consistently, and double-counting should be avoided. Contingency should be stated as a planned allowance rather than hidden padding.

Sample size should not be selected solely because a budget permits a particular number. It depends on the decision precision required, population variability, design, subgroup analysis, response quality, and nonresponse. For a simple proportion under a large-population approximation, a commonly used planning expression is:

Here, n₀ is the initial sample-size requirement, z is the critical value associated with the chosen confidence level, p is the anticipated population proportion, and e is the desired margin of error expressed as a proportion. The expression assumes a probability sample, an approximately independent sample, a proportion estimate, and the relevant large-sample approximation. If p is unknown, 0.50 produces the largest value of p(1−p) and is conservative for this purpose. Non-probability samples do not automatically justify a conventional margin-of-error claim, and clustering, weighting, design effects, finite populations, and nonresponse require further adjustment.

7. The Research Brief

A research brief is a shared agreement about why research is being conducted, what it must answer, how it will be conducted, and how findings will be used. It is not merely an administrative document.

A professional brief generally contains:

The brief should also identify known sources of evidence, such as analytics, customer-support records, prior studies, CRM data, market reports, and competitor information. Secondary data can refine the problem before primary data collection begins.

8. Methods Should Follow the Question

A common error is beginning with a preferred method: “We need a survey” or “Let us interview ten customers.” The method should follow the information requirement.

• Interviews and ethnographic observation are useful for meanings, experiences, workarounds, and unanticipated needs.
• Surveys are useful for estimating distributions, comparing defined groups, and measuring attitudes or intentions when sampling and measurement are credible.
• Usability studies reveal interaction problems and comprehension difficulties.
• Behavioural analytics describe observed journeys and changes over time but often cannot explain motives.
• Experiments and A/B tests can assess effects under controlled conditions.
• Social listening and review analysis may reveal naturally expressed concerns, but platform users are not necessarily representative of the market.
• Secondary research provides context efficiently but may use different definitions, dates, populations, or commercial interests.

Mixed methods are appropriate when one evidence type cannot answer the entire problem. For example, interviews may identify possible barriers, a survey may estimate their prevalence, and a usability test may evaluate solutions. The sequence should have a logic: later stages should use earlier findings rather than repeat them.

9. Illustrative Case: Reducing Abandonment in a Mobile Wallet

The following is a fictional illustrative case.

Background

A mobile-wallet company observed that registrations were growing, but only 41% of new users completed their first payment within seven days. The product team proposed shortening identity verification. Marketing believed that users did not understand the benefits, while compliance warned that verification could not be removed without regulatory review.

Problem

The initial request was “conduct a survey on onboarding.” This was reframed as a decision problem:

Which changes to onboarding should the company prioritise to increase first-payment activation without weakening required verification controls?

The research problem was that the relative contribution of comprehension, trust, perceived effort, technical failure, payment acceptance, and need for the product was unknown.

Objectives and questions

The primary objective was to identify and prioritise barriers to first payment among recently registered users. Secondary objectives were to compare barriers by acquisition source and user experience, and to evaluate alternative explanations for abandonment.

Questions included:

1. At what stage do users abandon onboarding?
2. What tasks and expectations do users bring to the process?
3. Which barriers are experienced as confusing, risky, time-consuming, or technically unsuccessful?
4. Do barriers differ between users who complete a first payment and those who do not?
5. Which proposed improvements increase comprehension and confidence while preserving compliance requirements?

Analysis and approach

The team first audited funnel data and support tickets. It found that completion of identity verification was not the only issue: many users completed verification but did not know where the wallet could be used. Interviews with recent users then revealed three distinct patterns: uncertainty about merchant acceptance, fear of making an irreversible mistake, and failure of document capture on low-bandwidth connections.

A structured survey was designed to estimate the prevalence of these barriers among eligible registrants. Because the company needed subgroup comparisons, the sample plan specified acquisition source, device type, and verification status in advance. A usability evaluation compared two compliant onboarding concepts: a revised explanation of acceptance locations and a guided first-payment flow. The evaluation measured task completion, comprehension, confidence, and observed errors. It did not claim to estimate long-term retention.

Outcome

The evidence led to a three-part intervention: clearer communication about where the wallet could be used, a low-bandwidth document-capture improvement, and a guided first payment. The company did not remove verification. The decision workshop ranked these actions by expected impact, implementation effort, compliance risk, and time to release.

Learning points

First, the original method request concealed several different information needs. Second, the decision context prevented an attractive but impermissible solution from being treated as the only option. Third, behavioural data, qualitative evidence, and evaluative testing answered different questions. Finally, the research did not “prove” one universal cause; it identified multiple barriers and supported prioritisation under constraints.

10. Common Framing Failures

Vague objectives

“Understand customer needs” provides no population, context, output, or decision link. It invites uncontrolled scope.

Solution-led research

Asking whether users like a predetermined feature can hide better alternatives. The brief should first establish the problem and decision alternatives.

Leading questions

Questions such as “How valuable is our innovative new service?” introduce claims and pressure respondents toward agreement.

Double-barrelled questions

“Was the app fast and easy to use?” asks about two constructs. A respondent may find it fast but difficult.

Confusing behaviour with intention

Stated willingness to purchase is not equivalent to purchase behaviour. Intentions can still be useful, but their limitations must be recognised.

Unbounded stakeholder lists

Trying to answer every stakeholder’s question in one study often produces a long instrument, weak attention, and shallow analysis.

Treating availability as representativeness

Employees, social-media followers, or the easiest customers to contact may systematically differ from the target population.

Overclaiming causality

Before-and-after changes may reflect seasonality, concurrent campaigns, product changes, or regression to the mean. The research design must match the strength of the claim.

11. Ethics, Governance, and Responsible Use of Data

Research participants should know the purpose of data collection, what participation involves, how information will be used, and whether participation is voluntary. Consent should be appropriate to the sensitivity and context of the research. Incentives should compensate fairly without becoming coercive.

Researchers must minimise collection of personally identifiable information, restrict access, define retention periods, and use secure storage. Sensitive data requires heightened safeguards and, where applicable, compliance with relevant privacy and consumer-protection law. Data should not be repurposed beyond the communicated scope without a justified basis and appropriate governance.

Product research can affect vulnerable users, including children, financially distressed consumers, patients, or workers whose participation may be monitored by employers. Recruitment and interpretation should avoid exploitation and stereotyping. Segment labels should describe evidence rather than stigmatise people.

AI tools can assist with drafting interview guides, clustering text, translating materials, summarising transcripts, and identifying possible patterns. They do not remove the need for human judgement. Researchers must check fabricated citations, distorted summaries, cultural bias, privacy leakage, automation bias, and the loss of contextual meaning. Sensitive transcripts should not be uploaded to an unapproved system. AI-generated hypotheses are prompts for investigation, not evidence.

12. Best-Practice Checklist

Before approving a study, ask:

1. What decision will the findings inform?
2. Who owns that decision and when must it be made?
3. What alternatives are genuinely open?
4. What uncertainty could change the decision?
5. Are the research problem and objectives distinct from the business problem?
6. Are questions neutral, specific, answerable, and prioritised?
7. Are hypotheses used only where testable and appropriate?
8. Are the population, unit of analysis, and time frame explicit?
9. Does the proposed method match the information requirement?
10. Are scope, budget, timeline, quality criteria, and exclusions documented?
11. Are claims proportionate to the design and evidence?
12. Are privacy, consent, inclusion, security, and AI governance addressed?

Chapter Summary

Effective market research begins with a decision, not a method. A business situation must be translated into a decision problem, then into a research problem that identifies the information gap preventing action. Clear objectives and research questions define what must be learned, while hypotheses provide testable propositions when quantitative or causal reasoning is appropriate.

The decision context includes stakeholders, alternatives, uncertainty, consequences, timing, and constraints. Information requirements operationalise abstract constructs into observable evidence. Scope, sample considerations, timelines, budgets, and deliverables make the study feasible and governable. A research brief creates shared agreement and protects the study from uncontrolled expansion.

Methods should follow the question. Qualitative, quantitative, behavioural, experimental, and secondary approaches each provide different forms of evidence, and mixed methods can combine their strengths. Researchers must avoid leading questions, solution-led framing, weak sampling, and causal overclaiming. Ethical governance, privacy protection, inclusion, and responsible use of AI are integral to professional research rather than optional additions.

Key Terms

Business situation: An observed organisational condition requiring attention.

Decision problem: The managerial choice or action that must be addressed.

Research problem: The information gap that prevents a sound decision.

Research question: A specific, answerable enquiry derived from the research problem.

Research objective: A statement of what the study will accomplish.

Hypothesis: A provisional, testable statement about a relationship or difference.

Decision owner: The person accountable for making the relevant decision.

Decision alternative: A realistic action, including maintaining the status quo.

Exploratory research: Research used to investigate meanings, experiences, and possible explanations.

Descriptive research: Research used to estimate characteristics, behaviours, or distributions.

Diagnostic research: Research examining possible reasons for an observed pattern.

Evaluative research: Research assessing the performance or likely value of an intervention or alternative.

Construct: An abstract concept such as trust, value, or perceived effort.

Operationalisation: The process of representing a construct through observable indicators.

Unit of analysis: The entity being studied, such as a person, account, session, or transaction.

Population: The full set of units to which conclusions are intended to apply.

Sample: The units actually included in a study.

Information requirement: A specific piece of knowledge needed to answer a question or support a decision.

Research brief: A document defining the purpose, scope, design, governance, schedule, and outputs of a study.

Confirmation bias: The tendency to seek or interpret evidence in ways that support an existing belief.

Review Questions

Short Answer Questions

1. Distinguish between a business problem and a research problem.
2. What is the role of a decision owner in a research project?
3. State two characteristics of a well-designed research question.
4. What is meant by an information requirement?
5. Why should a research brief specify exclusions as well as inclusions?

Long Answer Questions

1. Explain the chain from business situation to decision and show how each stage differs.
2. Discuss how uncertainty, consequences, and timing influence the value and design of market research.
3. Explain the relationship between research objectives, research questions, hypotheses, and measures.
4. Compare exploratory, descriptive, diagnostic, and evaluative research purposes, giving a product-management example of each.
5. Describe the contents of a professional research brief and explain how it supports research governance.

Analytical / Application Questions

1. A subscription product has high trial registrations but low conversion to paid plans. Develop a decision problem, one primary objective, four research questions, and three information requirements.
2. A product manager asks for “ten interviews to validate our new feature.” Critically assess this request and propose a more appropriate framing process.
3. Prepare a concise research brief for investigating why customers abandon an online checkout. Include the decision, scope, target population, proposed mixed-method design, timeline, deliverables, and ethical safeguards.

Multiple Choice Questions with Answers

Question 1

Which statement best describes a decision problem?

A. A broad desire to understand customers

B. A managerial choice or action that must be addressed

C. A list of survey questions

D. A summary of collected data

Correct Answer: B. A managerial choice or action that must be addressed

Question 2

Which objective is most appropriately written?

A. To understand everything about customers

B. To prove that the proposed feature is excellent

C. To identify barriers to first-week activation among newly registered users

D. To conduct ten interviews with users

Correct Answer: C. To identify barriers to first-week activation among newly registered users

Question 3

What is the main purpose of an information requirement?

A. To specify the knowledge needed to answer a question and support a decision

B. To determine the colour of a report

C. To replace the decision owner

D. To guarantee a causal conclusion

Correct Answer: A. To specify the knowledge needed to answer a question and support a decision

Question 4

Which research purpose is primarily concerned with estimating prevalence or frequency?

A. Exploratory

B. Descriptive

C. Ethnographic

D. Creative

Correct Answer: B. Descriptive

Question 5

Which statement about hypotheses is correct?

A. Every qualitative study must use formal hypotheses

B. A hypothesis is a provisional, testable statement

C. A hypothesis is the same as a business decision

D. A hypothesis guarantees causation

Correct Answer: B. A hypothesis is a provisional, testable statement

Question 6

Which item should normally be included in a research brief?

A. Only the final presentation design

B. Decision owner, objectives, scope, schedule, and governance

C. A guarantee that findings will support management’s preferred option

D. Only the number of participants

Correct Answer: B. Decision owner, objectives, scope, schedule, and governance

Question 7

Why is a solution-led research request risky?

A. It can prevent consideration of better alternatives

B. It always requires a larger sample

C. It eliminates the need for analysis

D. It guarantees representative data

Correct Answer: A. It can prevent consideration of better alternatives

Question 8

In the expression n₀ = z² × p(1 − p) ÷ e², what does e represent?

A. The population size

B. The critical value

C. The anticipated proportion

D. The desired margin of error

Correct Answer: D. The desired margin of error

Question 9

Which claim generally requires stronger design than a simple observational association?

A. Users reported confusion

B. Users who encountered an error contacted support

C. A redesigned interface caused retention to increase

D. Participants described trust concerns

Correct Answer: C. A redesigned interface caused retention to increase

Question 10

Which is an appropriate governance practice when using AI to summarise research transcripts?

A. Accept the summary without checking it

B. Upload all sensitive transcripts to any free tool

C. Verify outputs and use approved systems with privacy safeguards

D. Treat generated themes as established evidence

Correct Answer: C. Verify outputs and use approved systems with privacy safeguards