Use customer interviews when you need to understand how and why a decision unfolded, including triggers, alternatives, language and context. Use a survey when you have clear, testable questions and need structured responses from a larger defined population. Interviews generate and refine explanations; surveys can estimate patterns. Neither method is automatically stronger, and a staged combination is often more useful.
Match the method to the question
Choose interviews for questions such as:
- What caused this problem to become urgent?
- How did buyers handle it before?
- Which alternatives entered the decision?
- What does “easy” mean in this workflow?
- Why did an objection matter?
Choose surveys for questions such as:
- How many respondents in a defined customer population report a given behaviour?
- Which of several clearly written problems is most frequently selected?
- How does a measured response differ between defined groups?
- Has an established measure changed over time?
A survey is not a faster interview. It cannot probe an ambiguous answer or notice that your response options omit the real alternative. An interview is not a tiny survey. Repetition across a few conversations does not produce a reliable market percentage.
Compare the practical differences
| Dimension | Interviews | Surveys |
|---|---|---|
| Best for | context, mechanisms, language | structured measurement and comparison |
| Question style | open and adaptive | standardised and pretested |
| Sample | small, purposive | larger, defined population |
| Analysis | interpretation with traceable evidence | statistical and response-quality analysis |
| Main risk | interviewer influence and overgeneralisation | question bias, poor sampling and false precision |
GOV.UK research guidance notes that qualitative rounds often involve a small number of participants, while surveys and benchmarking require far more participants for clear findings. The exact survey sample depends on population, sampling, expected effects and analysis—not a single universal threshold.
Use interviews before a survey when the language is unclear
If you do not yet know the relevant alternatives or vocabulary, start with interviews. Ask about recent events using customer interview questions for better positioning. Synthesise the accounts and turn the resulting hypotheses into neutral survey questions.
Pilot the survey with people from the intended population. Ask them to explain what each question means and how they selected an answer. This exposes ambiguous wording and missing options before a larger launch.
A useful sequence is:
- define the decision;
- interview relevant participants;
- form hypotheses;
- design and pilot a survey;
- field it to an appropriate sample;
- interpret results with sampling and non-response limits;
- follow up surprising patterns qualitatively.
Illustrative example
Imagine five interviews suggest founders delay rewriting a homepage because internal stakeholders cannot agree on the priority customer. A survey could ask a defined population about the barriers encountered in their latest homepage project, with options informed by interviews and an open response.
If disagreement is frequently reported, that supports its prevalence in the sampled population—not a universal causal claim. Follow-up interviews could explain how the disagreement was resolved. This is an illustrative research design, not a study result.
Avoid common survey shortcuts
Do not ask respondents to rank a long list of desirable benefits without trade-offs; almost everything may look important. Avoid double questions such as “How clear and persuasive is our message?” Include “not applicable” or “none” where truthful. Randomise options where order could bias selection, and do not treat convenience-panel responses as representative of a market.
For interviews, avoid recruiting only advocates, asking leading questions or turning every repeated phrase into a finding. The round guidance in how many customer interviews you need explains why relevance and variation matter.
Limitations and trade-offs
Interviews provide depth but require skilled moderation and interpretation. Surveys can reach more people, but poor sampling or wording can produce precise-looking misinformation. Response rates and non-response bias matter; so does whether the intended audience can be reached at all.
Mixed methods cost more time and coordination. Use them when the decision value warrants it, not because a plan looks more rigorous with two methods.
Frequently asked questions
Can a one-question poll replace customer interviews?
No. It may capture a directional response to that question, but it cannot reconstruct the experience or reveal missing answer choices. Use it only for a suitably narrow purpose.
Should a survey come before interviews?
It can when the constructs and answer options are already well understood. When the subject or language is uncertain, interviews usually make the later survey more valid.
Can interviews validate a percentage?
No. Interviews can explain a pattern or reveal mechanisms. Estimating prevalence requires an appropriate quantitative design and sample.
Your useful next step
Write the decision and classify each unknown as “explore”, “explain”, “estimate” or “compare”. Use interviews for the first two and consider a survey for the latter two. Store both methods and their limitations together using the customer research repository framework.