Recruitment, live interviews and analysis solve different problems. Build the right research stack, then keep every coded excerpt connected to its actual source.
The useful question is which research stage needs support, rather than which product wins every category. Official sources checked October 5, 2026. These are reasoned fit recommendations, not hands-on performance rankings.
A practical starting stack: consider User Interviews for recruitment logistics, Lookback for moderated sessions, and Dovetail for imported research analysis. You may need fewer tools if your existing recruitment and interviewing process already works. Check the exact plan, consent and access requirements before committing.
User Interviews
Find and organize participants
Its researcher support covers targeting, screeners, scheduling and incentive distribution, with Recruit for its panel and Research Hub for your own users. Treat this as the recruitment layer; its help overview is not evidence that it independently conducts or analyzes your interviews.
Before buying: can you recruit your actual audience with a screener that verifies relevant experience, within your budget and geography?
Lookback
Run a moderated session
Lookback distinguishes LiveShare, with continuous screen sharing, from conversation-focused Interview sessions. On mobile, LiveShare requires the Participate app; Interview can run without it but does not provide mobile screen sharing. Both allow observers and recordings. Confirm the supported device and session mode before inviting someone.
Before buying: can one participant join on their actual phone, give consent and complete the recording or screen-share workflow you need?
Dovetail
Organize imported evidence
Dovetail documents importing recordings, documents, research notes and survey spreadsheets into projects. Highlights and tags can connect source material to searchable evidence. Check import limits, permissions and the intended plan; import support alone does not demonstrate analysis quality or guarantee an insight.
Before buying: can a colleague trace your proposed finding back to the exact source excerpt, with appropriate access?
A small workbench. A clear evidence trail.
Paste an anonymized transcript, select an exact passage, and apply a human code and note. Compare coded excerpts and source coverage without pretending a few interviews represent an entire population.
Demo statements are invented solely to demonstrate coding. They are not customer research, findings or quotations from real people. Your pasted text is processed locally in this tab. No transcription, AI inference, recruitment, recording or automated thematic analysis occurs. Avoid entering sensitive identifiers; exports contain the full source text.
The saved source is immutable so offsets remain traceable. Selection positions are JavaScript UTF-16 character offsets, with an exclusive end. If you need to change a source, remove it and add the corrected transcript; removal also removes its excerpts.
Add an anonymized transcript or load the clearly synthetic demo.
No coded evidence yet.
PapaParse 5.4.1 produces the actual CSV, including quoting and spreadsheet-formula escaping. JSON preserves exact text, source offsets, participant IDs, definitions and human notes. Markdown and CSV reflect the current code filter; full study JSON always contains everything. Counts are source IDs and excerpts in this study, not prevalence, statistical significance, saturation or validated demand. Reload starts a fresh tab; download JSON before leaving and restore it explicitly.
Keep the decision separate from the quote.
A quote supports an observation. A pattern needs comparison and context. A product decision still needs a reason, uncertainty and a next check.
Recruit for experience, not agreement
Start from the decision and the experience you need to understand. A relevant screener can ask about a recent behavior and context. Do not recruit only people willing to praise an idea. Record your recruitment constraints alongside the study.
Ask for a recent concrete episode
Ask what happened, what the person tried, what got in the way and what they did next. Avoid presenting a solution before exploring the current workflow. Your questions are part of the context, so preserve them with the transcript.
Code excerpts with counterexamples
Choose passages that show evidence for and against a proposed interpretation. Keep participant IDs consistent; two sources from one person should not inflate coverage. This workbench allows human notes and exact offsets, but cannot determine whether a code or interpretation is valid.
Share the finding with its limits
Write the observation, selected excerpts, relevant context, counterevidence and the decision it might inform. State what remains unknown. A frequent code in a small purposive sample is not a population estimate or a purchase forecast.
No. They serve distinct research stages. Use the stage comparison to locate your bottleneck and test the exact workflow. A small team may use existing calls and notes before adopting dedicated recording or repository software.
Does this tool analyze interviews automatically?
No. You select the text, choose a code, define it and write an interpretation. The tool maintains exact source links and counts recorded evidence. It does not transcribe, infer sentiment, invent themes or judge interview quality.
How are participant counts calculated?
For each human code, the tool counts unique participant IDs among that code's excerpts. The same ID appears once even if multiple sources or excerpts use it. These counts describe your imported study only.
Can the evidence be exported safely?
Exports include the text you enter, including full transcripts in JSON. Remove personal information before importing and review the exported files before sharing. CSV cells beginning with formula markers are escaped for spreadsheet use; that does not anonymize their contents.