Thematic analysis tools: coding, synthesis, or collection

Junu Yang · 2026-09-07

Choose a thematic analysis tool for the job in front of you: coding the text, synthesizing themes, or collecting the interviews and open-ended responses. The nine tools are grouped by that job. The list is unranked.

Coding means you tag passages and keep a trail. Synthesis means themes, clusters, or a repository from material you already have. Collection means you still need the interviews and open-ended responses, including follow-up, and then the themes.

What's the best software for thematic analysis of user interviews?

If you still need to run the interviews, Usercall runs them as AI-moderated voice interviews and codes the themes, from $89 a month for about 10 to 20 interviews. If you already have transcripts and need a formal record with a codebook and inter-rater checks, NVivo, ATLAS.ti or MAXQDA fit better. Each Usercall theme links back to its quotes. For a searchable home for past research, use Dovetail.

What a thematic analysis project holds

The software stores interviews, survey responses, support tickets, and other customer feedback. You tag passages, group codes into themes, compare segments, and export what you found.

A spreadsheet can hold the quotes. The project file holds the codebook, the memos, and a way to pull the tagged passages back out. Automated coding, clustering, summaries, and follow-up during collection are extra. Only some of these tools include them.

Coding: tag the text and keep a trail

Choose a coding tool when the work is a codebook someone else can follow.

NVivo

NVivo is for a formal qualitative project. You get coding hierarchies, memos, queries, framework matrices, and inter-rater checks. Someone has to maintain the project, or that record never gets written.

ATLAS.ti

ATLAS.ti covers coding, memos, networks, multimedia, and mixed methods in the same formal category. Open it when the project mixes interviews, documents, and visual material. The coding structure helps when someone keeps it current.

MAXQDA

MAXQDA is for thematic coding that has to sit beside survey variables, demographics, or other structured fields. You set that comparison up in the project. Skip it when there are no structured variables to attach to the themes.

Quirkos

Quirkos is a visual coding app with a simpler screen, which is why a small team can learn it. The trade is depth and scale.

Taguette

The self-hosted open-source version of Taguette is free. You can tag text, organize excerpts, and collaborate lightly. Plan for setup time, weaker support, fewer advanced features, and rougher collaboration. It fits a thesis, a pilot, or a side study.

Delve

Delve is a coding workspace for a solo researcher or a small team that wants a structured place to code. Cauliflower, in the synthesis section, is the tool for clustering and summaries. Keep those jobs separate.

Synthesis: themes from text you already have

Choose a synthesis tool when the interviews, tickets, or open-ended responses are already collected.

Dovetail

Dovetail is a research repository for product and design teams. It starts from research that already exists. Highlights, tags, and collaboration are how those teams find it again.

Cauliflower

Cauliflower is an AI thematic analysis tool for clustering, summarization, and pattern detection. It proposes themes from text you already have. Read the clusters before you treat them as findings.

Collection: interviews and open-ended responses, then themes

Choose a collection tool when you still need to gather the interviews and open-ended responses, and you need those conversations coded into themes.

Usercall

Usercall is for product teams working from interviews and open-ended survey responses. It runs AI-moderated follow-up during collection and does automatic theme coding on that material. For a formal record with framework matrices and inter-rater checks, use NVivo, ATLAS.ti, or MAXQDA. For a repository of research the company already has, use Dovetail.

Automated coding still needs a reader

Use automated coding to draft themes from a large set of open text, then read the draft. Use manual coding when you need a trail, a codebook you can defend, or a reliability check between two coders.

The review steps are in Automated vs Manual Coding: Fast Thematic Analysis at Scale.

Academic records and product decisions are different jobs

An academic or evaluation project usually needs documented coding logic, memos, an explicit codebook, and sometimes inter-rater evidence. NVivo, ATLAS.ti, and MAXQDA are built to keep that record.

A product team usually needs themes for a roadmap, a message, onboarding, or support. Usercall covers collection and coding for interviews and open-ended responses. Dovetail covers a repository the rest of the team can open.

Four questions that name the job

What is the job: coding, synthesis, or collection?

Coding is a codebook and a trail. Synthesis is themes or a repository from text you already have. Collection is interviews and open-ended responses you still have to gather, with follow-up, and then code. The stuck job should pick the tool.

What data do you have?

For interviews and open-ended survey responses, with AI-moderated follow-up and automatic theme coding, use Usercall. For documents, images, variables, and a reliability check, use NVivo, ATLAS.ti, or MAXQDA.

Who opens it every week?

A trained researcher can work in a heavier coding tool. Product managers, designers, and support leads need search. If they cannot find the findings, the work stays with one person.

What has to survive a switch?

Codebook structure, tagged excerpts, source metadata, inter-rater history, and repository links can flatten in a move. Export a sample project before you sign.

If you are still choosing a method, read 12 Qualitative Data Analysis Methods That Actually Work and Content Thematic Analysis: Exact 6 Steps to Code Data.

Three things that break in a move

Codes often arrive damaged. Parent-child structures collapse, memo links disappear, and segment references get messy enough that the old history is hard to trust.

Agreement checks are the second loss. The new tool can allow collaborative coding and still drop the old reconciliation notes. Archive that evidence first if you need it.

The third loss is a tool only one researcher can operate. The rest of the team needs to read, search, and act on the findings.

Grouped examples of coded feedback are on Thematic analysis examples (real user feedback).

How the nine tools split

Coding with a formal record: NVivo, ATLAS.ti, and MAXQDA. Simpler coding: Quirkos and Delve. Free self-hosted coding: the open-source version of Taguette. Synthesis from research you already have: Dovetail for the repository, Cauliflower for clustering, summaries, and pattern detection. Collection, with AI-moderated follow-up and automatic theme coding, for interviews and open-ended responses: Usercall.