Market Research

Start with a decision and a question

Market research is not a pile of charts or a ritual performed before launching a project. It is the purposeful collection and interpretation of information about a market in order to inform a decision. OpenStax describes it as collecting and analyzing information related to a target market, while the U.S. Small Business Administration (SBA) frames it as bringing together consumer behavior and economic trends to improve an idea. In practice, a research effort earns its place when it links a question, evidence, and a decision that could change.

For example, a neighborhood grocery-delivery team may need to decide whether to investigate late-evening pickup. Its questions could be: Which nearby areas contain potential users? Are households concentrated near the proposed pickup point? What times do people report as inconvenient? How common is that constraint among the defined group? What alternatives already exist? These are different questions. A local population count can describe context, but it cannot establish why people make a particular choice. A small set of direct responses can reveal possible reasons or routines, but it cannot automatically estimate prevalence in a neighborhood.

Write the decision before choosing a method. “Learn about our customers” is too broad to guide evidence. “Decide whether to spend another week studying evening pickup for commuters within two miles of campus” is narrower. It identifies a group, geography, decision horizon, and uncertainty. A research question should also make clear what result would matter. If the team will continue studying the idea when evidence suggests a sizeable, reachable group with an unmet timing constraint, it should say what counts as suggestive evidence and what remains unknown. This is not a formula for approving or rejecting a business. It is a discipline for avoiding data collection that cannot affect the next move.

Market research touches related topics without replacing them. Customer discovery concentrates on testing a focused segment-and-problem assumption. Customer interviews are one way to gather direct evidence, but their recruiting and conversation mechanics are a separate lesson. Competitor analysis examines particular alternatives and firms. Value-proposition work frames a proposed customer benefit. This lesson stays at the research-design level: define the question, choose evidence fit for that question, and report the boundary of what was learned.

Primary and secondary research answer different questions

Primary research creates new data to answer a defined question. It might include a survey, structured observation, experiment, or group discussion, depending on the question and ethical constraints. The decisive feature is not the tool but the purpose: the team is gathering information it did not previously have. Primary research can be especially useful when the question is specific to a local setting, a new behavior, or a proposed attribute that no existing dataset measures. It requires time, design choices, and careful interpretation.

Secondary research uses data, reports, records, or other information that another organization collected earlier. It often answers broad and quantifiable questions about population, demographics, incomes, employment, industries, locations, or trends. The SBA cautions that existing sources may not be specific to one audience even when they are useful context. A researcher should therefore inspect who collected the information, when, where, for what original purpose, how key terms were defined, and whether the population matches the current question. A national figure may be sound for the nation and still be weak evidence for one campus, neighborhood, or narrow customer group.

Official sources can be a sensible starting point for secondary research. Census Business Builder, for example, provides selected demographic and economic data by location and business type. NAICS is the federal classification system used for U.S. business-economy statistics, so it can help a researcher locate data whose industry category is stated rather than guessed. Neither tool creates evidence of a specific venture’s demand. They provide context that must be connected cautiously to the claim at hand.

Primary and secondary evidence are often complementary. Suppose a team is considering a pickup point near campus. Secondary data may help describe nearby households, income measures, and relevant business categories. A direct study may then investigate a narrow practical question, such as the timing or access constraints of a defined group. The first does not replace the second, and the second does not erase the first. Combining sources is useful only when each source is allowed to answer the question it can actually support.

Population, sampling frame, and sample set the boundary of a claim

A population is the full group a research claim is about: for example, all enrolled commuter students who attend evening classes at one campus during a stated term. A sampling frame is the practical list or route from which a sample can be reached: perhaps a mailing list, a course roster, a customer database, or people who see a posted link. A sample is the portion that actually provides data. These three groups are frequently mistaken for one another. A campus newsletter list is not automatically all commuter students; the people who choose to answer a link are not automatically representative of everyone on that list.

The sampling frame matters because it reveals who could have been included and who was missed. A survey posted only in one social-media group may miss people who do not use that group. A response-based sample can also differ from nonrespondents in ways that matter to the question. Wording, timing, incentives, and the measurement itself can introduce additional limits. A large number of responses does not repair a mismatch between the intended population and the people reached.

Sample size still matters, but in a narrower way. For estimates based on samples, the Census Bureau explains that larger samples generally have less sampling error, and margins of error communicate uncertainty associated with sampling. That does not turn a margin of error into a universal accuracy label. It does not include every possible source of error, such as incomplete coverage, nonresponse, unclear questions, incorrect recording, or a conclusion that exceeds the data. The AAPOR ethics code similarly says that claims of precision and applicability to a broader population must be warranted by the sampling frame and methods used.

Report the scope before stating a finding. “Among 26 voluntary respondents reached through a campus newsletter, 14 reported a late-evening pickup problem” is a bounded description. “More than half of all commuters need this service” is a population claim that the same convenience sample cannot establish. The first can justify a next research question; the second requires a design that can support generalization.

Interpret evidence honestly and handle people responsibly

Interpretation begins by separating what the evidence observed from what the team infers. An estimate should travel with its population, date, source, method, and relevant uncertainty. A pattern may be a promising signal, a challenge to an assumption, or a reason to investigate a subgroup. It is not automatically proof of demand, willingness to pay, a market-size estimate, or a forecast. When sources conflict, do not average them into confidence. Check whether they used different geographies, time periods, definitions, samples, or measures. The disagreement may identify the next useful question.

Research involving people also has ethical obligations. Participation should be voluntary, and people should not be misled about the purpose of the activity. If identifiable information is collected, researchers should be truthful about its use, restrict access, and follow applicable rules. The Federal Trade Commission’s business guidance emphasizes data minimization: do not collect or retain sensitive personal information without a legitimate need, keep what is needed only as long as necessary, protect it, and dispose of it properly. Exact legal duties vary by jurisdiction, organization, and data type, so this lesson is a general ethical baseline rather than legal advice.

A compact research record helps make conclusions reviewable: state the decision; write the research question; identify the population and the accessible sampling frame; name the sources and dates; describe how the information was obtained; distinguish observations from interpretations; and list limitations and remaining uncertainty. This record makes it possible for a later reader to challenge an inference without losing the underlying evidence. Market research is most useful when it makes both the next decision and the limits of confidence clearer.