Writing Effective Questions

The quality of your questions determines the quality of insights you'll uncover. Diaform supports open text, single choice, multiple choice, NPS, and star rating questions. Open text is still best for deeper discovery, while structured question types are useful when you need a clear selection or score.

Question Types#

Open Text#

Open text questions let respondents answer in their own words. Use them for discovery, complaints, feature requests, churn research, and anything where context matters.

Single Choice#

Single choice questions let respondents pick one answer option. Use them for segmentation, routing, or simple preference questions.

Add at least two answer options.

Multiple Choice#

Multiple choice questions let respondents select more than one answer option. Use them for "select all that apply" questions such as feature usage, purchase drivers, or pain-point lists.

Add at least two answer options.

NPS#

NPS questions can use the standard 0 to 10 scale or up to five custom options. Use NPS when you want a recommendation-intent score and still want the AI to ask follow-up questions about the reason behind the score.

Star Rating#

Star rating questions render a 1-to-N star input. You can set the maximum number of stars from 1 to 10.

Open-Ended vs Structured Questions#

Prefer open-ended questions for richer, more actionable data. Closed questions (yes/no, ratings, multiple choice) limit what respondents can tell you. Open-ended questions give the AI room to explore context, motivations, and nuance.

Keep It Conversational#

Write questions as if you're talking to a friend, not conducting an academic survey. Conversational questions feel natural and encourage honest, detailed responses.

Academic: "Please provide feedback regarding your product utilization experience."

Conversational: "What has your experience been like using our product?"

Recommended Question Count#

3-5 questions is ideal for most use cases. More questions mean longer interviews, which can lead to respondent fatigue and drop-off.

• 3 questions: Quick pulse checks, post-transaction feedback (5-10 min)
• 4-5 questions: Customer satisfaction, onboarding feedback (10-15 min)
• 6+ questions: Deep research, product discovery (15-25 min)

Remember: the AI asks multiple follow-ups per question, so even 3 questions can generate substantial depth.

Good vs Bad Examples#

Here are real-world examples that demonstrate the difference between surface-level and insight-generating questions:

Example 1: Product Experience#

Bad: "Rate our product 1-10"

Good: "What has your experience been like using our product?"

Why: The rating gives you a number but no context. The open-ended version invites stories, specific examples, and natural follow-up opportunities (e.g., "Can you tell me more about that frustration you mentioned?").

Use rating, NPS, or choice questions when you need structured data, then let the AI follow up to collect the reason behind the selection.

Example 2: Recommendation Intent#

Bad: "Would you recommend us?"

Good: "If a colleague asked about us, what would you tell them?"

Why: The yes/no version is binary. The alternative reveals what aspects they'd highlight, how they describe your value, and what reservations they might have.

Example 3: General Feedback#

Bad: "Any feedback?"

Good: "What's one thing you wish worked differently?"

Why: "Any feedback?" is too vague—most people will say "no" or give generic answers. Asking for one specific improvement makes it easier to respond and surfaces actionable insights.

Focus on Past Behavior Over Hypotheticals#

Ask about what people have actually done, not what they think they might do in the future. Past behavior is a more reliable indicator of needs, pain points, and decision-making.

Hypothetical: "Would you use a feature that does X?"

Behavior-focused: "Tell me about the last time you tried to do X. What did you end up doing?"

One Topic Per Question#

Don't combine multiple topics into a single question. It confuses respondents and makes it harder for the AI to follow up effectively.

Combined: "What do you like about our product and what would you improve?"

Separated:

• "What's working well for you in our product?"
• "What's one thing you wish worked differently?"

Question Limits Per Plan#

Your plan determines how many questions you can add per conversation:

• Trial / Starter: Up to 6 questions
• Standard: Up to 10 questions
• Pro: Up to 15 questions
• Business / Enterprise: Up to 20 questions

If you need more questions, consider splitting your research into multiple projects or upgrading your plan.

Required Questions#

Mark a question as required when the AI must collect that answer before moving on.

Required questions are useful for contact details, eligibility checks, or critical research inputs. For personal data such as email, name, phone, or date, the AI checks that the respondent provided the requested information.

If you want to collect an email address, add it as a required question instead of looking for an email collection setting in General Settings.

Next Steps#