Conduct Primary Market Research

Proof required

Submit: a research plan document covering research question, method justification, sampling strategy, and sample size rationale; the pilot-tested instrument (survey or interview guide) showing the pre-pilot and post-pilot versions; and a pilot log documenting the changes made and the reason for each.

What gets checked

• Research question is specific and testable — 'why do customers churn in month 3?' passes; 'understand customer behaviour' fails
• Sample size is justified for the chosen method — not an arbitrary number
• Pilot changes are documented: at least two questions were revised based on confusion, ambiguity, or leading phrasing

Common mistakes

• Skipping the pilot entirely and collecting data with an unvalidated instrument that produces unusable responses
• Writing leading questions that confirm a pre-existing hypothesis rather than testing it

Resources

What a verifier looks for

• Read three questions from the instrument and ask the candidate to identify which assumption each tests
• Ask what changed between the pre-pilot and post-pilot versions and why
• Challenge the sample size: ask the candidate to justify the number without referring to convention
• Check for leading questions: pick two and ask whether a respondent could easily answer them without actually holding the relevant opinion

You'll sign in first, then come straight back here.

Proof required

Submit: the raw data file (with personally identifiable information removed or anonymised); the cleaned dataset; a data cleaning log documenting every transformation decision (duplicate removal, outlier treatment, recoding) and the reason for each; and a fieldwork summary noting recruitment method, response rate, and any sample quality issues.

What gets checked

• At least 30 survey responses or 8 depth interview transcripts are present in the raw data file
• Every cleaning decision is documented — no silent deletions or transformations
• The fieldwork summary acknowledges at least one sample quality limitation honestly

Common mistakes

• Convenience sampling from friends and colleagues without acknowledging the resulting sample bias
• Cleaning data selectively — removing inconvenient responses without documenting the rationale

Resources

What a verifier looks for

• Ask to see the raw data alongside the cleaned data — check that the cleaning log accounts for every row removed
• Probe the recruitment method: ask exactly how respondents were found and whether they represent the intended sample
• Challenge one outlier treatment decision: why was that response removed rather than kept?
• Ask whether the response rate affected the analysis and how the candidate accounts for potential non-response bias

You'll sign in first, then come straight back here.

Proof required

Submit: the full research report (four to six pages) covering methodology, findings with supporting evidence, limitations section acknowledging at least two constraints, and recommendations; and a written record of the defence session including the reviewer's credentials, the specific challenges raised, and your responses.

What gets checked

• Findings are presented with appropriate confidence language — 'the data suggests' rather than 'the data proves' for small samples
• At least one counter-interpretation of the main finding is acknowledged and addressed in the report
• Reviewer has research credentials (academic or professional) and posed at least one methodological challenge beyond scope or sample size

Common mistakes

• Reporting only findings that confirm the original hypothesis and omitting contradictory data
• Seeking a reviewer who is unfamiliar with research methodology — the defence is meaningless without someone who can challenge analytical choices

Resources

What a verifier looks for

• Ask the candidate to identify their main finding and then propose an alternative explanation for the same data
• Check the limitations section: is it specific and honest, or vague and boilerplate?
• Probe one recommendation: ask what evidence in the data specifically supports that recommendation
• Challenge the sample: ask how confident they are that the findings generalise beyond the people surveyed

You'll sign in first, then come straight back here.