Better Questions, Better Research: Using AI to Improve Research Materials
And the silver medal goes to...
Making research materials better is one of the best uses of AI in healthcare market research.
Before we interview a single patient or physician, our questions are already shaping what we will learn.
A leading question can introduce bias. An unclear instruction can confuse respondents. An overlooked topic can leave an important business question unanswered.
AI can help us identify these weaknesses before fieldwork begins.
Turning a brief into a stronger starting point
Imagine a client wants to understand why physicians are reluctant to prescribe a recently launched treatment.
AI can help translate that brief into an initial discussion guide, suggesting areas to explore: familiarity, clinical confidence, patient suitability, practical barriers and previous prescribing experiences.
The researcher then challenges those suggestions against the business objectives and existing knowledge.
Importantly, the guide should allow respondents to explain their decisions without assuming that reluctance exists or prescribing should increase.
Spotting questions that steer the answer
Consider this question:
“What benefits have you seen since switching patients to the new treatment?”
It assumes there have been benefits.
AI can flag that assumption and suggest a more neutral alternative:
“What changes, if any, have you observed since switching patients to this treatment?”
That small adjustment gives respondents room to describe improvements, disadvantages or no change.
AI can act as a useful second reader, although researchers still need to assess whether its suggested revisions are appropriate.
Developing probes that go further
A patient says their treatment is “inconvenient”. What should we ask next?
AI can suggest probes exploring appointment frequency, travel, administration, storage or disruption to work and family life.
These suggestions can help moderators prepare for different possibilities.
The skill during the interview is choosing which direction to follow, based on what that person actually says. A prepared list should support listening, rather than dictate the conversation.
Improving questionnaires and screeners
AI can also review draft questionnaires for repetition, ambiguous wording and response options that overlap.
For example, age bands of “30–40” and “40–50” leave a 40-year-old uncertain which to select.
It can flag potential routing problems or inconsistencies between screening criteria and later questions.
However, reviewing a document does not prove that a programmed survey works. Researchers must still test the actual routes, eligibility rules and respondent experience.
Making materials easier to understand
Healthcare language can become unnecessarily complicated.
AI can suggest clearer alternatives to technical wording or help adapt materials for patients, caregivers and healthcare professionals.
Those revisions need checking to preserve clinical meaning and avoid introducing assumptions about literacy or understanding.
Where confidential briefs or unpublished product information are involved, only approved AI services with verified data protections should be used.
The researcher remains the author
For me, AI’s greatest contribution here is giving us another opportunity to question our questions.
It can generate alternatives and expose weaknesses. Experienced researchers decide what belongs in the final materials.
The result should be a conversation that serves the research objectives while giving respondents space to surprise us.