Results
Browse responses in two views:- Transcript view — a list of interviews with the full conversation transcript alongside. Voice and video interviews include an audio player, video playback, and word-level timestamps.
- Grid view — a spreadsheet with one row per respondent and one column per question, ideal for scanning structured answers.
Per-interview AI analysis
Every completed interview gets automatic AI analysis:- Summary — an overview of the conversation.
- Sentiment — positive, neutral, or negative, with the reasoning.
- Quality score — a rating with explanation.
- Fraud detection — automatic flagging of spam and low-quality responses, with reasoning.
Insights
The Insights tab analyzes the whole study:- Study summary — an executive overview with key bullets.
- Per-question insights — themes, a summary, and sentiment for each question, with response counts.
- Charts — auto-generated visualizations for quantitative questions (multiple choice, rating scales, NPS, star ratings, numbers), including bar charts and heatmaps for matrix questions.
- Custom insights — define your own AI analysis prompts (“What pricing objections came up?”) and get reusable, reorderable insight views.
Quotes with receipts
Every insight is backed by evidence: quotes link to the exact interview, position in the transcript, and timestamp — with one click to jump to the source.Analysis controls
- Analysis mode — Basic (faster, cost-effective) or Pro (deeper analysis).
- Regenerate — refresh the summary, a single insight, or everything after new responses arrive.
- Scope — include or exclude incomplete transcripts and simulated interviews.
Filtering
Filter both Results and Insights by:- Question answers (equals, includes, greater/less than) on quantitative types
- Metadata fields (including hidden fields from your links)
- Interview status — completed, archived, screened out
- Simulated vs. real interviews
- Split-test arm — compare variants in studies with a randomizer
Reports
Build cross-study reports (documents or presentations) with an AI chat assistant: pull in findings from multiple studies, add custom analysis, and share the result publicly with a revocable link. Reports track whether their source studies have new data since the report was written.Exports
- PDF — formatted reports with a cover page, executive summary, custom insights, and per-question pages with charts and data tables. Optimized for large studies (1,000+ responses).
- CSV — raw response data, sanitized for Excel compatibility. Multi-select multiple choice answers export as one column per option (cell filled when selected, blank when not, plus an Other column when free-text entries are enabled) — the same fan-out matrix, ranking, and allocation questions use — so per-option pivots work directly in Excel. An allocation question with custom options turned on adds two last columns: Custom options (total), the sum of the values a participant gave to their own options, and Custom options, their names and values as text. The two columns are exported only while the setting is on.
Telling repeated questions apart
Concept tests often repeat the same questions once per stimulus — three concept blocks inside one randomized group, say — so the question text alone can’t tell you which concept a column belongs to. The CSV export adds that context to each column header:- Group titles — a question inside a titled group is prefixed with the group’s breadcrumb:
3. Concepts › Concept B › How appealing is this?. Untitled groups add nothing, so name your concept blocks to get labeled columns. - Attached media — a question with media attached lists the file names after its text:
1. Take a look at this [concept-b.png]. - Order shown — each randomized group gets an extra
<group title> (order shown)column recording the order that participant actually met its contents in, pipe-separated so it splits cleanly:Concept B|Concept C|Concept A. Blocks the participant never reached are left out. An untitled block is named by the first file attached inside it.