Creating an interview
After you create a study, most quality comes from clear objectives and session shape before publish. Use the dynamic builder and AI co-pilot so you are not writing a full discussion guide by hand.
Create the study
- From the dashboard or Insight list, create a new interview (default title is Untitled / Başlıksız).
- You land in the dynamic builder: chat with the assistant on one side; objectives and form fields update as you go.
- Templates can pre-fill objectives for common use cases (onboarding, NPS, CSAT, etc.).
Start with one primary goal
Ask: “What decision will this conversation inform?” Write one primary goal — for example “Understand why trial users churn before connecting their data.” Vague goals (“get feedback”) produce vague interviews. Use the in-builder AI assistant to tighten wording.
Objectives panel
- Primary goal & target insight — what “done” looks like for the report.
- Respondent persona — who is talking (trial user, buyer, employee).
- Tone of voice — friendly check-in vs formal research.
- Information goals / themes — what the agent should explore after required fields.
- Prohibited topics — hard off-limits during dialogue.
- Session shape — pulse vs deep dive, open vs guided. Controls follow-up depth after answers; do not put minute caps here.
Session shape (important)
Session shape tells the agent how to behave across formats without rewriting prompts: a quick pulse with one follow-up, a form-heavy slot-by-slot flow, or a deep dive with several probes on pain points. Change session shape when the same brief needs a different pacing.
AI co-pilot in the builder
Describe what you want in plain language. The assistant can draft objectives and structured form rows (including types and options). You review, edit labels, and save — the agent does not publish for you.
What the AI needs to know (context)
In the Context tab, add briefing items and attachments (links, images) with notes for the agent. These can be shown mid-chat when relevant. Also add approved answers for meta questions (privacy, response times) so the agent does not invent policy.
Participant appearance & voice
- Welcome copy, logo, theme, and background for the public chat.
- Voice chat mode — text only, voice, or let participants choose. Voice uses AI read-aloud and mic input; credits differ from text.
- Privacy policy link — participants can consent before chat starts.
Common mistakes
- Too many competing goals in one interview — split studies if needed.
- Internal jargon in objectives participants will not understand.
- Skipping draft test preview (?test=1) before sharing widely.
- No session shape — agent pacing feels random across sessions.
- Putting clock limits (“6 minutes”) into session shape — use depth/style language instead.