4.1 Compiling and evaluating data collected

Once fieldwork ends, a marketing team holds a pile of questionnaires, interview notes, sales records and observation sheets that cannot yet answer any business question. Compiling means gathering this raw material into one organised, clean and usable form. Evaluating means examining that organised data to judge its quality, summarise what it shows and decide what it means for the organisation. Example: a supermarket chain that has collected 400 shopper questionnaires must first check, code and tabulate them before it can say which promotions customers value most. The steps in this topic prepare the findings that are later turned into recommendations and a written report.

Reviewing and editing raw data

Editing is the first check on collected data. Its purpose is to find omissions, unclear entries and contradictions while the fieldwork is still fresh enough to correct them. A questionnaire that is skipped at this stage can distort every figure that follows. Example: a county revenue office surveying market traders finds that some forms record a trader's monthly income but leave out the type of stall, so the forms cannot be placed in the correct group.

Points to check when editing:

• Completeness: Confirm that every question has an answer or a clear note that it was not applicable. Forms with large gaps are set aside or followed up.
• Legibility: Entries that cannot be read, such as unclear handwriting on a paper form, are clarified with the enumerator who collected them.
• Consistency: Answers must not contradict each other. A respondent who says they have never visited a hotel but then rates its breakfast has made an error that must be resolved.
• Accuracy of units: Amounts, dates and quantities must use the same units. A farm survey that mixes kilograms and 90 kg bags needs conversion to one unit.
• Uniformity: Responses written in different forms, such as "Nrb", "Nairobi" and "NAIROBI", are made to follow one agreed format.
• Authenticity: Forms that look copied, filled in a single pattern or completed by one hand under many names are flagged as possible fabricated returns.

Coding and entering data

Coding assigns a number or short label to each possible answer so that responses can be counted quickly and entered into a spreadsheet or statistical package. Closed questions are pre-coded on the form, while open responses are coded after reading a sample of answers and grouping similar ones. Example: a bank surveying account holders codes "Which channel do you use most?" as 1 = branch, 2 = mobile app, 3 = agent, 4 = ATM.

Procedure for coding and entering data:

1. Prepare a codebook. List every question, its variable name, the possible answers and the code for each. The codebook keeps all team members coding in the same way.
2. Number each questionnaire. Give every form a unique serial number so that any entry can be traced back to the original document when an error is suspected.
3. Code open-ended answers. Read a sample of responses, group similar ideas into categories, assign codes and add a code for "other" and for "no response".
4. Set up the data sheet. Create a spreadsheet where each row is one respondent and each column is one variable, with clear column headings matching the codebook.
5. Enter the data. Type the codes carefully, preferably with two people sharing the work, one reading and one entering.
6. Verify the entry. Re-enter or check a random sample, for example every tenth form, against the originals and correct any mismatch.
7. Back up the file. Save a copy in a second location and keep the original forms safe until the analysis is complete.

Tools commonly used in compiling data:

Cleaning and organising data

Cleaning removes or corrects errors that remain after editing and entry. Unclean data produces misleading averages and wrong percentages, so cleaning is done before any analysis. Example: a hospital's patient satisfaction file contains an age of 250 for one respondent, which is clearly a typing error that would inflate the average age if left uncorrected.

Common cleaning tasks:

• Removing duplicates: The same questionnaire entered twice is deleted so that one respondent is not counted twice.
• Handling missing values: Blanks are marked with an agreed code, and the analyst decides whether to exclude those cases from a question or to follow up the respondent.
• Detecting outliers: Values far from the rest, such as a monthly spend of Ksh 900,000 among shoppers who mostly spend under Ksh 10,000, are checked against the original form. If genuine, they are kept but reported separately.
• Checking ranges: Each variable has a valid range. A rating scale of 1 to 5 should contain no 7, and any such value is traced and corrected.
• Standardising categories: Spelling variants and similar answers are merged under one label so that counts are not split.
• Documenting changes: Every correction is recorded so that another person can see what was changed and why.

Tabulating and presenting data

Tabulation arranges the coded data into tables that show how often each answer occurred. A frequency table is the simplest form, and it forms the basis for charts and for comparison between groups. Example: an agrovet shop asked 200 customers how they usually pay for farm inputs.

Percentage = (number in category ÷ total respondents) × 100

For mobile money:

Percentage = (110 ÷ 200) × 100

Percentage = 0.55 × 100

Percentage = 55%

Every table or chart carries a clear title, labelled axes or columns, units and the sample size. Ways of presenting tabulated results:

• Frequency tables: Show counts and percentages for one question and are the starting point for all other displays.
• Cross-tabulation: Splits one variable by another, for example preferred payment method by age group, to reveal differences between segments.
• Bar charts: Compare categories of equal standing, such as the number of customers choosing each payment method.
• Pie charts: Show how parts make up a whole and work best with few categories that add to 100%.
• Line graphs: Show change over time, such as monthly sales across a year.
• Histograms: Show how numerical values such as customer age or spend are spread across ranges.

Analysing quantitative data

Quantitative analysis uses numbers to summarise and compare. The measures of central tendency describe the typical response, while percentages and rates allow comparison between groups of different sizes. Choosing the right measure matters: the mean is sensitive to extreme values, while the median is not. Example: a hotel collected ten guest ratings on a scale of 1 to 5.

Ratings: 4, 5, 3, 4, 2, 5, 4, 3, 4, 5

Mean = sum of values ÷ number of values

Sum = 4 + 5 + 3 + 4 + 2 + 5 + 4 + 3 + 4 + 5 = 39

Mean = 39 ÷ 10 = 3.9

Median: arrange in order 2, 3, 3, 4, 4, 4, 4, 5, 5, 5. The 5th and 6th values are both 4, so the median = 4.

Mode: the value that appears most often is 4.

Other useful calculations:

• Percentage: A SACCO surveyed 250 members and 85 said they prefer applying for loans on their phones.
  • Percentage = (number in category ÷ total) × 100
  • Percentage = (85 ÷ 250) × 100
  • Percentage = 34%
• Response rate: A school sent 240 questionnaires to parents and received 180 back.
  • Response rate = (questionnaires returned ÷ questionnaires sent) × 100
  • Response rate = (180 ÷ 240) × 100
  • Response rate = 75%
• Percentage change: A bakery's monthly sales rose from Ksh 480,000 to Ksh 540,000.
  • Change = ((new value − old value) ÷ old value) × 100
  • Change = ((540,000 − 480,000) ÷ 480,000) × 100
  • Change = (60,000 ÷ 480,000) × 100
  • Change = 12.5% increase
• Range: The highest value minus the lowest value shows the spread. In the hotel ratings, the range is 5 − 2 = 3.
• Market share: A firm's sales divided by total market sales, multiplied by 100, shows its position against competitors. A bakery with sales of Ksh 3,000,000 in a town market worth Ksh 20,000,000 has a share of (3,000,000 ÷ 20,000,000) × 100 = 15%.

Analysing qualitative data

Qualitative data comes from interviews, focus groups, open questions and observation notes. It explains why customers behave as they do, and it supplies the reasons behind the numbers. The task is to organise words into themes without losing their meaning. Example: a contractor's firm interviews ten clients about delays on completed projects and finds that "poor communication" appears in most interviews even though no question asked about it directly.

Procedure for thematic analysis:

1. Transcribe or write up. Convert recordings and notes into text soon after the session so that detail is not lost.
2. Read through the whole set. Read every transcript to get a general sense of what respondents said before marking anything.
3. Generate initial codes. Mark short labels beside phrases that carry a distinct idea, such as "late delivery" or "friendly staff".
4. Group codes into themes. Combine related codes into broader themes, for example "service reliability".
5. Review the themes. Check each theme against the original text to make sure it is supported and does not overlap with another theme.
6. Select supporting quotations. Choose short, representative quotes that illustrate each theme.
7. Count where useful. Note how many participants raised each theme to show how widespread it is, without presenting small groups as statistically representative.

Evaluating the quality of the data

Evaluation asks whether the data can be trusted before conclusions are drawn. Poor data leads to poor decisions, however neatly it is analysed. Example: a farm cooperative surveying only members who attend monthly meetings cannot claim the results represent all members, because those who rarely attend were left out.

Criteria for judging data quality:

• Validity: The data must measure what it claims to measure. A question about "satisfaction" that actually asks about price does not give valid results.
• Reliability: Repeating the study under similar conditions should give similar results. Inconsistent answers across similar questions suggest low reliability.
• Representativeness: The sample should reflect the target market in age, location, income and other relevant features. Check how respondents were chosen.
• Adequate sample size and response rate: Very small samples or low response rates weaken confidence in the findings.
• Bias: Look for leading questions, interviewer influence, non-response bias and respondents giving answers they believe are expected.
• Timeliness: Old data may no longer describe current customer behaviour, particularly in fast-changing markets such as mobile services.
• Source credibility: Secondary data from other organisations should come from a known, reputable source with a clear method.
• Triangulation: Compare findings from different sources, such as survey results against sales records, to see whether they agree.

Interpreting findings and stating limitations

Interpretation turns figures and themes into meaning for the organisation. The analyst compares results against the research objectives, against earlier periods or competitors, and against what was expected. Example: if 34% of SACCO members prefer phone-based loan applications, the analyst asks whether that share is rising, which age groups it comes from and what it means for the service channels members use.

Steps in sound interpretation:

1. Return to the objectives. State which research question each finding answers.
2. Identify patterns and differences. Note the largest and smallest categories, changes over time and differences between segments.
3. Look for reasons. Use qualitative findings and market knowledge to explain why the pattern appears.
4. Separate fact from opinion. Present what the data shows, and mark clearly any inference that goes beyond it.
5. Avoid claiming cause too easily. Two measures moving together do not prove that one causes the other.
6. State the limitations. Record weaknesses such as small sample, non-response, short time period or data gaps, so that readers judge the findings fairly.
7. Record the key findings. List the main results in plain language with supporting figures, ready to be used when recommendations and the report are prepared.

Chapter Summary

This chapter focused on the crucial steps involved in preparing a market analysis report. It began with compiling and evaluating the collected data, emphasizing the importance of organizing information systematically and assessing its relevance and reliability to ensure accurate insights. The chapter then explored how to make well-founded recommendations based on the evaluated data, highlighting the need to align suggestions with market trends and organizational goals. Finally, it addressed the preparation of the marketing analysis report itself, detailing how to present findings clearly and professionally to support decision-making. The report must be structured logically and include key elements such as an executive summary, data analysis, conclusions, and actionable recommendations. Throughout the chapter, the emphasis was on accuracy, clarity, and practical value to guide effective market strategies.