Customer research workshop / 5 MIN READ
Ask one survey question at a time
Separate the things a customer might judge differently before turning answers into a percentage.
Original editorial learning framework
Editorial and source review: 6 November 2026
A neat score can hide an ambiguous question. If customers can reasonably agree with one part and disagree with another, the response cannot tell you which problem to fix.
What you will leave with: An edited question set with a clear decision attached to every item.
Use this when
- A customer survey combines several features in one rating.
- Stakeholders interpret the same score as support for different changes.
This is a questionnaire review exercise, not a guarantee of representative sampling or a substitute for specialist survey design.
Spot the bundled judgment
Pew Research Center cautions against questions that ask people to judge multiple concepts at once. Its guidance also emphasizes clear wording. The workflow below applies those principles to a marketing team deciding what to improve.
Take the draft question and underline each thing the respondent must evaluate. “How helpful and trustworthy was the product guide?” contains two judgments. Someone can find a guide useful while doubting a particular claim. A single score cannot reveal that distinction. Mark the likely decision beside each judgment: revise organization, add evidence, remove an unsupported statement, or investigate something else.
Separate only what changes a decision
Splitting a sentence does not justify doubling the questionnaire indefinitely. Keep an item only when a plausible answer would change what the team does. If every answer leads to the same preselected redesign, the survey is decorative. Ask the decision owner to write the alternatives before adding another question.
Avoid replacing a bundled question with a list of vague adjectives. “Useful,” “relevant” and “helpful” may generate several numbers without clearer diagnosis. Prefer a defined task: could the person find whether the product fits their device? Separate task completion from confidence in the compatibility claim.
Make the response fit the experience
A person who never opened the guide cannot fairly rate its usefulness. Add an eligibility question or an appropriate not-used option rather than forcing a midpoint. Define the reference experience, such as the most recent purchase, so different respondents are not silently judging different periods.
Review the scale with the intended audience. The labels must mean something in the actual language and context, not only to the research team. Ask a pilot participant to explain what they thought each question meant. Record ambiguity rather than coaching them toward the intended interpretation.
Keep the revision visible in reporting
If you replace a combined score with separate items, do not draw a continuous trend line as if the measure stayed identical. Label the change and preserve the earlier wording in the research archive. A new answer may reflect a better question rather than a change in customers.
Present the separate results beside the decision they inform, along with who answered and who was excluded. Where useful, retain an optional explanation prompt to understand an unexpected answer. Do not expose identifiable comments in a widely shared report. A better instrument should produce a clearer next action, not merely more charts.
Make the reasoning concrete.
A fictional appliance brand asks whether its installation instructions are “clear and complete.” A buyer understands every sentence but cannot find the required clearance behind the appliance.
Diagnosis
The combined rating cannot distinguish confusing language from missing information. The team writes separate task-based questions about finding the clearance and understanding the installation sequence.
Pilot the revised questions before sending them. Review missing-information answers with the technical owner; do not turn a survey response into an unverified installation recommendation.
Take the questions into your next brief.
Use these prompts in your own notes. Nothing is submitted from this page.
Decision
Which revision could this answer change?
Bundled concepts
Which parts could a respondent judge differently?
Eligibility
Who actually experienced what is being rated?
Answer options
Can someone say not used, not applicable or uncertain where appropriate?
Version record
How will the question change be disclosed in later comparisons?
Watch for the wrong conclusion.
- Splitting every adjective into a separate item without a decision.
- Treating a forced midpoint as neutral sentiment.
- Comparing old and new wording as one unchanged metric.
Your next action
Review five questions from your next survey. Remove or split each item that cannot support one interpretable decision.
Sources & scope
The teaching, hypothetical exercises and recommended workflows are original editorial analysis. Linked sources support only the scoped facts identified beside them; they do not validate predicted results or endorse this framework.
- Pew Research Center: Writing Survey Questions →Primary research-method guidance · Supports: Question wording, answer format and sequence can influence answers; the marketing exercises here are original editorial applications, not Pew findings.