4.4 Market research

Market research gathers and analyses information about customers, competitors and the market to reduce uncertainty in decisions. A sound process defines the question, chooses evidence, collects it ethically, analyses it and links findings to action.

Research has value only when the question and method fit the decision. Biased wording, weak samples or outdated data can create false confidence.

Start with the decision and information gap, then check source quality, sample fit and how the result will change the plan.

Before launching a meal-delivery service, a business tests which delivery times customers value and what price they would actually pay, rather than asking only whether the idea sounds good.

Research does not eliminate uncertainty; it improves a decision when its limits are explicit.

Primary research collects new data directly for the business’s current question, using methods such as surveys, interviews, observation or product tests.

It can be specific and current but costs time and money and may suffer from non-response or interviewer bias. The method should match the behaviour or opinion being measured.

Ask whether the data is new, directly collected and fit for the decision; then examine who was reached and who was missed.

A café observes queue times and interviews recent customers before changing its ordering system; this is primary evidence tailored to the decision.

Primary does not mean automatically accurate; poor design can make new data misleading.

Match each authorized primary method to the information need: surveys collect standardized answers efficiently from many respondents; interviews probe individual reasons in depth; focus groups reveal interaction, language and reactions among a moderated group; observations record actual behavior rather than claimed behavior. Surveys can be shallow, interviews and focus groups costly or moderator-sensitive, and observations cannot always reveal motives.

Secondary research uses data already collected by another party or for another purpose, such as government statistics, trade reports, company filings or internal records.

It is faster and cheaper, but definitions, dates, incentives and methods may not match the current question. Triangulating sources improves confidence.

Check who collected it, when, for what population and purpose, then decide whether it is comparable.

A retailer uses census income data to screen locations, then checks whether the data is current and measured at the same geographic level.

A published number is not neutral or necessarily relevant; source fit matters more than apparent authority.

Apply the specified secondary sources deliberately: market analyses summarize industries and competitors; academic journals provide researched concepts or findings but may be technical or slow to publish; government publications can offer large official datasets but may use broad or outdated categories; media articles are current but vary in evidence and incentives; online content is fast and broad but requires checks of authorship, date, method and purpose.

Qualitative research explores meanings, reasons and experiences through words or observation; quantitative research measures frequencies, amounts or relationships numerically. They answer different questions and can complement each other.

Qualitative depth can reveal an unanticipated need but is harder to generalise; quantitative breadth supports comparison but may miss why behaviour occurs. Combining them can test both pattern and explanation.

Match the method to the question: “how many/how often?” points to quantitative evidence; “why/how?” to qualitative evidence.

A survey finds 60% abandon an app at payment; follow-up interviews reveal confusing fees as the reason.

A percentage does not explain motivation, and a few vivid interviews do not estimate market size.

A sample is a selected subset of a target population. Random sampling gives each member a known equal chance of selection; quota sampling fills predefined category proportions; convenience sampling selects people who are easiest to reach. They trade representativeness, cost, speed and access.

Sampling error and bias arise when some groups are over- or under-represented, when people self-select or when the sampling frame excludes likely customers. Larger samples reduce random error but not systematic bias.

Define the population and selection rule, then ask who had a chance to be included and whether the sample mirrors the target.

To study commuters, a random sample requires a usable passenger list or selection frame; a quota sample can ensure specified proportions by travel time or age; a convenience survey taken only at 9 a.m. is quick but may miss night-shift workers and other groups.

A large convenience sample can be less reliable than a smaller, well-designed sample.