Glossary

What is thematic analysis?

Thematic analysis is the method of reading qualitative data (interviews, open-text survey answers, call transcripts) and tagging recurring ideas.

The tags are then grouped into themes that describe what the whole body of conversations is actually about.

The traditional process is: get familiar with the data, generate initial codes, group codes into candidate themes, review and merge them, name them, and write up with supporting quotes. AI tooling now does the first three steps well and the last two badly, because naming a theme is a judgment about what matters.

Why it matters

Without it, qualitative research collapses into whichever quotes the analyst remembers. Thematic analysis is the difference between "several people mentioned onboarding" and a defensible count of how many, who, and in what terms.

Example

Across 22 interviews, 47 coded excerpts cluster into five themes. The largest, 14 excerpts from 11 people, is first named "reporting problems," but reading the quotes shows it is really about explaining results to someone else. The rename changes the roadmap implication.

In UpSight

UpSight extracts the quotes first, then groups them into themes across conversations. Each theme links to the quotes it is built from and to the people who said them, with a count of each. Open the theme and read the quotes before you trust the label. The naming judgment stays with you.

Related terms

FAQ

What is thematic analysis in simple terms?
Reading everything customers said, marking the ideas that keep coming up, and grouping them into a short list of what the research is about.
Can AI do thematic analysis?
AI is strong at coding and clustering excerpts and weak at judging which theme matters. Treat the output as a first pass to review against the quotes.
How many themes should emerge?
Most studies land on four to seven meaningful themes. Twenty themes usually means the codes were never merged.