Interpreting data and making decisions from market research – Market Research (Senior 5 Entrepreneurship) | Ganzaa

Market Research

Lessons

Senior 5 · Entrepreneurship

Market Research

4 lessons

Lesson 4 of 4

Interpreting data and making decisions from market research

25 min

Sixty forms on the table

Kayitesi's survey is finished: 60 completed questionnaires from the stratified sample (15 bank staff, 21 government staff and 24 shop workers). A pile of paper decides nothing. To be useful, the answers must be checked, counted, summarised, compared and, above all, interpreted: what do they mean for the decision she has to make?

From raw answers to a decision

Edit: check every form; set aside incomplete or contradictory ones.
Code and tabulate: give each answer a code and count how often it occurs in a frequency table.
Summarise: calculate percentages and averages (mean, median, mode).
Present: draw tables and charts that make patterns visible.
Compare groups: cross-tabulate the answers by type of customer.
Judge reliability: how big and how fair was the sample; how far can it be trusted?
Decide and recommend: link each finding to an action, and state the risk that remains.

Definition

Frequency table: a table showing how many times each answer or value occurs in a set of data.

Example: "0 days: 24 people; 1–2 days: 21 people" is the start of a frequency table.

Definition

Cross-tabulation: a table that shows how the answers to one question differ between groups of respondents, for example by job, age or area.

Example: the share of bank staff, government staff and shop workers who would order delivered lunch.

Worked example 1: summarising one question

Question: "If a hot lunch were delivered within 30 minutes for Frw 2,000, how many days a week would you order it?"

Answer	Frequency	Percentage	Mid-value (days)	Frequency × mid-value
0 days	24	40.0%	0	0
1–2 days	21	35.0%	1.5	31.5
3–4 days	11	18.3%	3.5	38.5
5 days	4	6.7%	5	20
Total	60	100%		90

Percentages: for example $\frac{24}{60} \times 100\% = 40.0\%$ and $\frac{11}{60} \times 100\% = 18.3\%$ .

Mean days per worker: $\frac{90}{60} = 1.5$ days a week. The mode (most common answer) is 0 days, and the median (the 30th and 31st answers in order) falls in "1–2 days".

Worked example 2: comparing groups

Of the 36 respondents who would order on at least one day, 13 are bank staff, 12 government staff and 11 shop workers.

Group	Would order (at least 1 day)	Asked	Share of the group
Bank staff	13	15	$\frac{13}{15} \times 100\% = 86.7\%$
Government staff	12	21	$\frac{12}{21} \times 100\% = 57.1\%$
Shop workers	11	24	$\frac{11}{24} \times 100\% = 45.8\%$
All	36	60	60.0%

Interpretation: the overall figure (60%) hides a big difference. Bank staff are almost twice as keen as shop workers, probably because they have higher pay and cannot leave their desks for long. Kayitesi's first marketing should target banks and offices; shop workers may need a cheaper, smaller meal.

From findings to a decision

A survey measures what people say they will do. Experience in many markets is that only a fraction of stated intentions turn into real purchases, so careful researchers discount them. Kayitesi decides to assume that only one in five of the stated orders will happen, and compares the result with her break-even point (S4 taught that break-even output is fixed costs divided by contribution per unit).

Worked example 3: should Kayitesi go ahead?

Fixed costs: Frw 1,200,000 a month. Price Frw 2,000 and variable cost Frw 1,100 a meal, so the contribution is $2,000 - 1,100 = 900$ Frw a meal. (Illustrative figures.)

Break-even: $\frac{1{,}200{,}000}{900} = 1{,}333.3$ , so 1,334 meals a month, or $\frac{1{,}334}{22} \approx 61$ meals a working day.

Estimate A (market share, lesson 2): a 5% share gives 54 meals a day. Monthly result: $54 \times 22 \times 900 - 1{,}200{,}000 = 1{,}069{,}200 - 1{,}200{,}000 = -130{,}800$ Frw, a loss.

Estimate B (survey intention): $1{,}800 \times 1.5 \times \frac{1}{5} = 540$ orders a week, or $\frac{540}{5} = 108$ a day. Monthly result: $108 \times 22 \times 900 - 1{,}200{,}000 = 2{,}138{,}400 - 1{,}200{,}000 = 938{,}400$ Frw, a profit.

Decision: the two estimates fall on opposite sides of break-even, so the evidence is not yet strong enough to commit Frw 2.4 million. A sensible recommendation is to (1) run a two-week test market from a shared kitchen with a few bank branches, which costs little and measures real orders, and (2) look for ways to cut fixed costs so that break-even falls below the cautious estimate. Research has not said "yes" or "no"; it has shown which risk remains and the cheapest way to reduce it.

How far can findings be trusted?

Ask	Why it matters
Was the sample large enough?	Results from 60 people vary more by chance than results from 600.
Was it representative?	A sample of one group, or of volunteers, misleads about the whole market.
Were the questions neutral?	Leading questions inflate "yes" answers.
Is it up to date?	Prices, incomes and competitors change.
Do different sources agree?	Findings confirmed by two methods (here, market share and survey) are more reliable.

tutor2u, "Sampling in Marketing Research". Watch for why results from a sample are only estimates, and how the size and choice of the sample affect how far findings can be trusted.

Warning

Percentages can hide small numbers. "86.7% of bank staff" sounds strong, but it is 13 people out of 15. Always report the number asked beside a percentage, and be cautious about conclusions drawn from small groups.

Tip

The "so what?" test. After every finding, write "So …" and the action it implies. "40% would never order. So I must not plan for the whole market; I should target the 60% who would." A finding with no "so" is not yet an interpretation.

✦

Practice

List the steps in turning raw survey answers into a decision.
In a survey of 80 shoppers, 28 chose Brand X, 20 Brand Y, 22 Brand Z and 10 no preference. Draw a frequency table with percentages and name the mode.
A survey of 50 people asks how many litres of milk they buy a week: 0 litres, 10 people; 2 litres, 15 people; 4 litres, 20 people; 6 litres, 5 people. Calculate the mean.
Of 40 women asked, 26 would buy a product; of 60 men asked, 21 would. Cross-tabulate and interpret.
A bakery's fixed costs are Frw 900,000 a month and each loaf gives a contribution of Frw 300. Research suggests sales of 2,500 to 3,500 loaves a month. Calculate break-even and advise the owner.
Give three reasons why the result of a survey may be unreliable.
"75% of the people we asked said they would buy our new yoghurt, so we should produce for 75% of the town." Evaluate this conclusion.

Show answersHide answers

Edit; code and tabulate; summarise (percentages, averages); present (tables, charts); compare groups (cross-tabulate); judge reliability; decide and recommend.
Brand X 28 (35.0%); Brand Y 20 (25.0%); Brand Z 22 (27.5%); no preference 10 (12.5%); total 80 (100%). Mode: Brand X.
$\frac{10 \times 0 + 15 \times 2 + 20 \times 4 + 5 \times 6}{50} = \frac{0 + 30 + 80 + 30}{50} = \frac{140}{50} = 2.8$ litres a week.
Women: $\frac{26}{40} \times 100\% = 65\%$ ; men: $\frac{21}{60} \times 100\% = 35\%$ ; all: $\frac{47}{100} = 47\%$ . Women are much keener, so promotion should target women first.
Break-even $= \frac{900{,}000}{300} = 3{,}000$ loaves a month. The research range (2,500–3,500) straddles break-even, so the bakery could make a loss or a profit; advise testing the market, cutting fixed costs or raising the contribution before committing heavily.
Any three: the sample was small; it was not representative; questions were leading or unclear; respondents said what they thought was wanted; the data are out of date; the interviewer influenced answers.
The conclusion is too optimistic. Stated intentions overstate real purchases; the sample may not represent the whole town (were they all shoppers at one supermarket?); "would buy" does not say how often or how much; and competitors will react. The firm should discount the figure, check the sample, and test-market before producing on that scale.

1Sixty forms on the table

2From raw answers to a decision

3From findings to a decision

4How far can findings be trusted?

Sixty forms on the table

From raw answers to a decision

From findings to a decision

How far can findings be trusted?