Tone and paralinguistics
How something was said: hesitation, emphasis, sarcasm, the pause before a price. Measured across every session, not guessed from a quote.
Analysis and insights
Read every interview.
Not just the transcript.
Themes, tone, emotion and behaviour, analysed across every session, checked by trained annotators, and delivered as a report you can defend.
A transcript tells you what people said. Cookiy tells you what they meant.
Automated insights that go past word clouds: interview analysis on the recording itself, thematic analysis across the whole study, sentiment and emotion analysis on the timeline, and human review so the findings hold up.
How analysis works
Four timelines.
One finding.
When the interviews end, the work starts. Each conversation is broken into four aligned timelines, then analysed by models and reviewed by people.
P14 · Checkout study18:02 → 21:40
TranscriptWho said what, and when
TonePitch, pace, pauses, hesitation
EmotionFrustration, doubt, delight
BehaviourSkips, rereads, what they did on screen
“It's fine, I guess… [pause 2.1s] I just didn't expect the extra fee.”
Human labelling
Trained annotators code themes, check the AI's tags and label against your codebook. Linguists and domain experts when the study needs them.
Offline, across studies
Re-run analysis on past interviews with a new question. Compare cohorts, markets and waves without going back to field.
From raw sessions to decisions
What the analysis produces.
Coded themes
Every finding traced to the quotes and clips behind it, with how many participants said it.
Signal markers
Hesitation, frustration, delight and drop-offs marked where they happened.
Segment comparisons
Cohorts, markets and personas side by side, from interviews and surveys.
Your codebook
Annotators label against your framework when you have one.
Re-analysis on demand
New question, same interviews. Re-code past studies without new fieldwork.
Exports
Transcripts, coded data and clips to your repository or slides.
Deliverables
Ready to share
the moment fieldwork ends.
Report
- Background
- Business question
- Respondent profile
- Executive summary
- Key takeaways
- Recommendations
Written findings, ranked
A structured report with every claim linked to the quote and clip behind it.
Slides34%drop off at workspace setup
A deck your stakeholders read
Charts and numbers built from the interviews and surveys. No manual slide work.
Clip · 2:45
“A wall of logos.”
P8, DevOps lead, at 20:08The moments that matter
Key quotes and hesitations clipped with timestamp, tags and transcript.
Keep exploring
Related: Voice and tone analysis, Live observation and Qualitative research.
About analysis.
What does Cookiy analyse in an interview?
Four aligned timelines: what was said (transcript), how it was said (tone, pace, pauses), how the person felt (emotion) and what they did (screen and behaviour). Themes are coded across all sessions and ranked.
Is the analysis done by AI or by people?
Both. Models code themes and mark signals across every session, and trained annotators check the tags and label against your codebook when a study needs it. Linguists and domain experts join for specialist studies.
Can I re-analyse old interviews?
Yes. Ask a new question of past studies and Cookiy re-codes the existing interviews, so you can compare cohorts, markets and waves without going back to field.
What deliverables do I get?
A written report with ranked findings, stakeholder-ready slides with charts, highlight clips with timestamps and tags, and exports of transcripts and coded data.
How is this different from sentiment analysis tools?
Sentiment tools score text. Cookiy works on the recording itself, aligns tone and behaviour with the words, and has people review what matters, so findings hold up in front of stakeholders.
How fast is the analysis?
Findings update as interviews finish. Most studies have first findings within about four hours and a full report within 24.
Skip the synthesis backlog.
Free to start. No credit card. Video on this site is licensed stock footage of real people.