How to get great results 

‍

AI Chat works best when your questions are clear, focused, and grounded in the context of your study. The more detail you include, the better AI Chat can target exactly what you’re after and the more useful the output will be.

‍

At a high level: 

‍

• Be specific.
• Ask for examples, not summaries.
• Request quotes.

‍

Recommendations: 

‍

• Ask specific questions about a feature, process, or steps, e.g. “checkout,” “onboarding,” “pricing”
• Try adding context e.g. “from mobile users,” “during setup”
• Request quotes or examples. Try prompts like “Show me quotes…” to get quotes or examples.
• Ask about patterns, not single mentions.
• Focus on one topic per thread to keep responses precise and easy to build on.
• Customize a deliverable.

‍

Avoid: 

‍

• Vague or broad questions e.g. "Tell me about UX." or "Did users like it?”
• Internal jargon
• Asking multiple unrelated questions. Open a new chat as a separate thread for exploring new topics.
• General summaries e.g. "Summarize the feedback"
• Follow-ups unrelated to the original topic or previous messages.  
• Simple yes/no. "Did users like it?" AI Chat will answer this but we recommend gathering additional evidence for to answer this with specific citations.

‍

Example Question Prompts

‍

Here are a few examples to use as inspiration when forming your own questions.

‍

Pain Points

• What specific frustrations did users mention about checkout?
• Show me examples where users hit blockers during sign-up.

‍

Quotes & Language

• Show me memorable quotes about pricing.
• What language did participants use when describing trust?

‍

Feature Feedback

• What reactions did users have to the dashboard?

‍

Common Query Patterns

• Show me... (Evidence Gathering)
• What did users say about... (Thematic Exploration)
• How did users describe... (Understanding Mental Models)
• What patterns emerged around... (Theme Detection)
• What recurring complaints... (Theme Detection)

‍

Refining Queries

‍

Too broad
"What did users say about the product?".
This works but is not guaranteed to uncover the complete picture

↓
Add specificity
‍"What did users say about dashboard navigation?"

↓
Request examples
‍"Show me examples where users got confused in the dashboard"