Turn qualitative data into defensible themes.

Yes — it's what the tool is built for. Use literature review mode for your review chapter, pasting the findings of each study you're synthesising, or primary research mode for your own interview, focus group, and open-ended survey data. Either way you get named codes, a verbatim quote behind every code, themes with the sources that support and contradict them, and a Word report you can work from. Two things to do before you submit: review and rename the codes yourself — AI-assisted coding is a first pass, not a finished analysis — and check your institution's policy on AI assistance, disclosing it in your methods chapter where required. Cite the framework you chose (Braun & Clarke, Thomas & Harden, and so on), not this tool, as your method.

Substantially. A student NVivo licence runs about $125 a year and past $400 with the AI add-on, before the week you'd spend learning it. Here the first 3 sources of every analysis are free, and paid passes start at just $7 for up to 20 sources, or $12 for an advanced 50-source pass, with no word limit per source. A twenty-study review chapter costs $7, nothing to install, and you're running your first analysis within a few minutes. Check your university library first, though: a site licence sometimes makes NVivo free for you.

In literature review mode: upload each paper as you downloaded it (PDF or Word, up to 3MB free or 10MB on a paid pass). The engine keeps what a critical review needs, the methods, findings, discussion and conclusion, and leaves out the abstract, introduction, literature review and references, then reads the authors and year for the citation label. You don't need to trim anything; only the study's own findings are coded, never the studies it cites. In primary research mode: your own participant data, whether every interview is in one file or each participant has their own, plus transcripts, focus groups and survey exports (Word, PDF, TXT, CSV or Excel). A file holding several participants is split into one card each. Everything is read in your browser, and if a file can't be read (a scanned PDF, a password, an old .doc), you're told which file, what's wrong and how to fix it.

Refunds are available within 7 days of buying a pass if an analysis fails and still fails when you retry it. A failed run isn't counted against your pass, so retrying costs nothing. To request one, email hello@thematicanalysis.ai with the email address you paid with, the pass you bought, and the error message you saw. Qualifying refunds go back to your original payment method and usually take about 7 days to arrive, and the pass's access ends when the refund is issued. Refunds aren't given for completed analyses, files that couldn't be imported, or a change of mind. The full details are on our Refund Policy page.

Yes — switch to primary research mode on the analyse page. Paste or upload each participant's responses (interview transcripts, focus group excerpts, or open-ended survey answers), pick a framework, and the engine codes every participant with verbatim quotes and clusters the codes into themes across your dataset. Your first 3 participants are free; a full analysis of up to 20 sources starts at $7, or up to 50 for $12.

Five: Reflexive Thematic Analysis (Braun & Clarke) — the default and the most widely taught; Thematic Synthesis (Thomas & Harden) for systematic-review style syntheses; Framework Analysis (Ritchie & Spencer); Grounded Theory with constant comparison; and Qualitative Content Analysis. The engine follows the phases and terminology of whichever you pick, and the report names the method so you can cite it.

The first 3 sources of every analysis are free — studies or participants, no signup, no card, up to 15,000 characters (~2,500 words) in each box — for up to 2 analysis runs. Analysing more than 3 requires a paid pass starting at $7 for up to 20 sources, or $12 for up to 50 sources: the same price whether you run 4 sources or the maximum, no subscription, and the per-source word limit is lifted to around 20,000 words so whole papers and full transcripts fit. Paying unlocks your chosen pass for 30 days: the $7 Thesis Sprint Pass includes 5 full analysis runs, and the $12 Advanced Project Pass includes 10 runs plus 3 generations each of the Demographics Table, Research Gap Finder and Theoretical Underpinnings, and 7 Ask Your Data questions. A run that fails isn't counted. No account, no card stored.

Yes — this is one. NVivo runs about $125 a year on a student licence and north of $400 once the AI add-on is included; here the first 3 studies of every analysis are free and paid passes start at just $7 for up to 20 sources, so a ten-study synthesis costs $7 with nothing to install and no week spent learning the software. Other affordable options for qualitative analysis are Taguette (free and open source, manual coding only), Quirkos (~$21 a quarter) and ATLAS.ti (~$50–99 a year on a student desktop licence). Check your university library first, though: a site licence often makes NVivo free for you. And if you're hand-coding hundreds of primary transcripts for a funded study with a formal audit trail, NVivo or ATLAS.ti is still the right tool — this is built to get you from a pile of findings to defensible themes fast.

Yes — as an AI-assisted first pass, not a replacement for your judgement. Review every code against its quote, rename themes to fit your research question, and follow your institution's or journal's policy on disclosing AI assistance. The report includes a limitations note that says exactly this.

Every code carries a verbatim quote copied from the text you pasted, and every theme lists the studies and codes behind it — so each claim is checkable in seconds. The engine is instructed never to paraphrase or invent quotes, and anything you can't verify against your own sources, you should cut. That audit trail is the difference between this and asking a chatbot.

Your text is processed to produce your results and is not stored on our servers — analyses live in your browser session and disappear when you leave (so export your report before closing the tab). To run the analysis your text is sent to a third-party AI provider whose API we use; we choose providers that do not use API data to train their models, and may change provider over time. Anonymise participant data before analysing it.

No. Open the tool, paste your studies, run the analysis, download the report. Nothing to sign up for, nothing to cancel.

At least 3 — below that, cross-study themes aren't meaningful — up to 20 sources on the Thesis Sprint Pass and up to 50 sources on the Advanced Project Pass, with 10 or 25 extra swaps for fixing or replacing sources. Re-running sources you've already analysed never counts again. If you have more than 50 sources, start with your 50 most central studies, or analyse each sub-topic as its own project.