FIELD GUIDE 06 / Foundations
Market research: ask better questions before buying media
Research is useful when it reduces uncertainty around a real decision. The method should follow the question: interviews explain language and motives, observation reveals behaviour, surveys estimate patterns and experiments test causes.
QUICK ANSWER
The idea in one minute.
How interviews, observation, surveys and behavioural data answer different questions. Start with a decision, not a questionnaire Use more than one kind of evidence Document uncertainty and sampling limits
KEY IDEAS
What you will learn
- Start with a decision, not a questionnaire
- Use more than one kind of evidence
- Document uncertainty and sampling limits
01
Define the decision first
A research brief should name the decision, the people affected, what is already known and what evidence would change the plan. ‘Which feature should lead the launch message?’ is actionable. ‘What do people think about our brand?’ is too broad until the team defines the choice it needs to make.
Separate exploration from measurement. Early conversations can surface vocabulary, barriers and unexpected alternatives. A later survey can estimate how common those themes are. Treating a handful of interviews as population statistics, or using a large survey to discover questions nobody thought to ask, weakens the result.
02
Match the method to the evidence
Interviews are strong for stories and explanations but sensitive to memory and social desirability. Observation shows what happens but may not explain why. Transaction, search and service data record behaviour in a particular system; they do not automatically represent the whole market. Experiments can isolate causes, but only inside the conditions tested.
Triangulation compares evidence produced in different ways. If interviews describe a checkout barrier, analytics show abandonment at the same step and a usability test reveals the confusion, the conclusion is stronger than any one signal. Disagreement is also useful because it shows where assumptions need another test.
03
Design a survey people can answer
Use neutral wording, one idea per question and response options that cover realistic answers. Ask about a defined period rather than ‘usually’, randomise option order when order could influence choice and test the questionnaire with a small group before launch. Every extra question creates fatigue and should earn its place.
Sampling matters more than a polished chart. Record who was invited, who responded, how respondents differ from the target population and how much uncertainty surrounds each estimate. A precise percentage from an unrepresentative sample is still an unreliable guide to the market.
04
Turn findings into a decision log
A useful report connects observation, interpretation and action. State what the evidence shows, what the team infers, what remains unknown and which decision follows. This prevents a memorable quotation or colourful graph from becoming a claim the research never supported.
Keep the raw questionnaire, field dates, sample rules, analysis choices and important limitations. Future teams can then compare results honestly instead of repeating the same study. Research creates organisational memory when its reasoning is as visible as its conclusion.
SOURCE DESK
Research and further reading
These links lead to public guidance, open textbooks or freely accessible research. This guide explains the ideas in original language; open the sources to examine context, definitions and limitations.
- The Marketing Research Process ↗OpenStax introduction to defining a problem, collecting data, analysing findings and reporting results.
- Primary Market Research Methods ↗OpenStax comparison of observation, interviews, focus groups and surveys.
- Writing Survey Questions ↗Pew Research Center guidance on wording, order and response effects.
- Questionnaire design ↗UK Government Analysis Function guidance on clear questions and respondent burden.