Market Research
Senior 5 · Entrepreneurship
Market Research
Lesson 3 of 4
25 min
Kayitesi's first draft questionnaire began like this:
1. Don't you agree that delivered lunch would save you time?
2. How much do you earn and how much do you spend on food?
3. Would you like tasty, cheap and healthy meals brought to your desk?
Her friend Habimana, who studies statistics at university, read it and laughed: "Nobody will say no to question 3, and nobody will answer question 2 honestly." The tool you choose, the people you ask and the way you word each question decide whether research tells you the truth or only what you hoped to hear.
Definition
Research tool (method): the technique used to collect market information, such as a questionnaire, an interview, observation, a focus group or a test market.
Example: standing at a street corner and counting how many workers carry food back to their offices is the tool called observation.
| Tool | How it works | Strengths | Weaknesses |
|---|---|---|---|
| Questionnaire survey | The same written questions put to many people, face to face, by phone or online (a web form or a WhatsApp link) | Reaches many people cheaply; answers are easy to count | Low response rates online; badly worded questions give misleading answers |
| Personal interview | A one-to-one conversation, structured (fixed questions) or unstructured (open discussion) | Rich, detailed answers; the interviewer can ask follow-up questions | Slow and costly per person; the interviewer can influence answers |
| Observation | Watching and recording behaviour: counting passers-by, timing queues, "mystery shopping" at a rival | Records what people do, not what they say | Does not explain why; people may act differently if they know they are watched |
| Focus group | A guided discussion with about 6–10 target customers | Reveals attitudes and ideas; good for testing a new product concept | Small, not representative; one loud member can sway the rest |
| Test marketing | Selling the product for a short time in a small area before a full launch | Shows real buying behaviour with real money | Costs money; competitors see the idea early |
| Consumer panel | The same group of people reports its buying regularly over months | Shows trends and changes over time | Expensive to keep; members may drop out or change habits |
Two Teachers, "Primary Market Research Explained". Watch for the four methods in the title (surveys, focus groups, observations and test marketing) and for when each one is the right choice.
Kayitesi cannot question all 1,800 workers in the area. She questions a sample and uses the results to estimate what the whole group thinks. A sample is only useful if it represents the group.
Definition
Population (in research): the whole group the business wants to learn about. Sample: the smaller group actually questioned. Sampling frame: the list from which the sample is drawn.
Example: population, the 1,800 workers in the area; sampling frame, a list of the offices and shops with their staff numbers; sample, the 60 workers interviewed.
| Sampling method | How the sample is chosen | Main strength or weakness |
|---|---|---|
| Simple random | Every member of the population has an equal chance, e.g. names drawn by lottery or random numbers | Free of the researcher's bias, but needs a full list |
| Systematic | Every person on the list or passing a point, e.g. every 10th customer | Simple to run; biased if the list has a pattern |
| Stratified | The population is divided into groups (strata) and each group is sampled in proportion to its size | Represents every group fairly |
| Quota | The interviewer must find a set number from each group (e.g. 20 women, 20 men) but chooses whom | Quick and cheap; the interviewer's choice can bias it |
| Cluster | A few areas (clusters) are chosen at random and everyone in them is surveyed | Cheap when people are spread widely; clusters may differ |
| Convenience | Whoever is easiest to reach, such as friends or passers-by | Fast, but least representative |
Worked example 1: a stratified sample
The 1,800 workers are 450 in banks, 630 in government offices and 720 in shops. Kayitesi will interview 60. How many should come from each group?
Step 1: each group's share of the population. Banks ; government ; shops .
Step 2: apply each share to the sample of 60. Banks ; government ; shops .
Check: . Within each group she then chooses people at random, for instance every third name on each office's staff list.
Worked example 2: repairing bad questions
| Draft question | Fault | Better version |
|---|---|---|
| Don't you agree that delivered lunch would save you time? | Leading: it pushes towards "yes" | How do you usually get your lunch on a working day? (tick one) |
| How much do you earn and how much do you spend on food? | Double-barrelled (two questions in one) and too personal | On a normal working day, how much do you spend on lunch? Under Frw 1,000 / Frw 1,000–1,999 / Frw 2,000–2,999 / Frw 3,000 or more |
| Would you like tasty, cheap and healthy meals brought to your desk? | Loaded words; everyone says yes, so it measures nothing | If a hot lunch were delivered within 30 minutes for Frw 2,000, how many days a week would you order it? 0 / 1–2 / 3–4 / 5 |
The repaired questions ask about behaviour and real prices, use answer bands that do not overlap, and give no hint of the "right" answer.
Warning
Respect the people you survey. Tell them who you are and why you are asking, ask their consent, collect no more personal data than you need and keep it safe. Rwanda's Law N° 058 of 2021 relating to the protection of personal data and privacy governs how personal data such as names and phone numbers may be collected and used.
Tip
Match the tool to the question. "How many?" needs counting: a survey or observation. "Why?" needs talking: interviews or a focus group. "Will they really pay?" needs real money: a test market.
Practice
Sources and credits
Written by Ganzaa: the questionnaire case, the tables, the sampling diagram, the worked examples and the practice items. The business, names and figures are illustrative (2026).
