What is thematic analysis?
Thematic analysis finds the patterns in written feedback by tagging each answer with the topics it mentions and grouping those tags into themes. Counting the themes shows what customers raise most, and following them over time shows what is growing.
From many comments to a few themes
Open answers are hard to read in bulk. Thematic analysis turns them into something you can count: every answer gets one or more short tags, called codes, such as “late parcel” or “tracking link broken”, and related codes are grouped into a theme, such as delivery.
The result is a short list of themes, each with a count and a few quotes. It tells a team which problems are common, which are rare but serious and which are new this month, and it gives a score its cause.
The method comes from qualitative research, where it is used on interviews, and it works the same way on survey answers, support tickets and reviews. For feedback that arrives every week, a fixed list of themes keeps this month's counts comparable with last month's.
- Works on
- Open survey answers, interviews, support tickets and reviews
- Produces
- Themes with counts, trends and example quotes
- Steps
- Read, code, group into themes, count, review, report
- Two approaches
- Inductive, with themes from the answers, or deductive, with a set list
How to do thematic analysis
Work through the answers in six steps. The first round takes longest, and once the list of themes settles, new answers can be coded against it quickly.
Agree on the list with everyone who codes. Two people coding the same answers differently produce counts that can't be trusted.
- Read a sample of answers to get a feel for what people say
- Give each answer short codes for the topics it mentions
- Group related codes into themes, such as delivery, price or login
- Write one line describing each theme, so anyone codes it the same way
- Count the answers in each theme, period by period
- Report the top themes with their counts, trends and a few quotes
A thematic analysis example
Say 250 customers answer “What could we have done better?” in a month. After coding, the answers fall into three main themes and a group that fits none of them.
Delivery leads with 32% of the answers. If it held 20% the month before, it is the theme to look at first, even when prices draw the angriest comments.
- Delivery times
- 80 answers, 32%
- Prices
- 55 answers, 22%
- Checkout
- 40 answers, 16%
- Other or no theme
- 75 answers, 30%
- Top theme
- Delivery, 32%
How to calculate a theme's share
A count shows size, and a share lets you compare periods with different numbers of answers. Report both, along with the change from the period before.
Divide the answers in a theme by all the answers you read in the same period, including the ones that fit no theme.
Theme share = answers in the theme ÷ answers read × 100
An answer that mentions two topics can count in two themes, so the shares can add up to more than 100%. Say so when you report them.
Inductive vs deductive thematic analysis
Inductive analysis builds the themes from the answers themselves, which suits a new topic. Deductive analysis files answers into a list decided in advance, which suits tracking the same topics over months.
A common approach combines them: build a list inductively from a first batch, code new answers against it, and review the list whenever the group with no theme starts to grow.
| Inductive | Deductive | |
|---|---|---|
| Themes come from | The answers themselves | A list set in advance |
| Best for | Exploring a new topic | Tracking known topics over time |
| Strength | Finds what nobody expected | Counts stay comparable between periods |
| Risk | Themes shift as you read more | New topics land in “Other” |
How Usertick groups feedback into themes
On every plan, Usertick files each written answer under a category and a narrower theme from your organization's own list as the response arrives. The list starts from a set for your industry chosen at signup and changes only when an admin edits it, so counts stay comparable from one period to the next. When the AI is unsure of a category or theme, the field stays blank instead of guessing.
On every plan, Settings → Analysis lets you add categories and themes, edit what each one should catch, or describe a new one and let AI draft it, from 5 categories on Free to 30 on Business. Before you save, Usertick shows how many of your last 100 written answers the draft would catch, with examples. Changes, renames included, apply to new responses.
Results lists the survey's themes with their counts and change from the period before, marking new ones, and on hover shows an example answer and the country where the theme is most over-represented. Anyone with permission to manage responses can move an answer to another theme, and the correction counts everywhere.
Mistakes to avoid with thematic analysis
Small choices that keep the results useful.
- 01
Changing the theme list every month
A new list breaks the comparison with last month. Change themes on purpose and note when you did.
- 02
Themes that are too broad
A theme such as “Product” holds everything and explains nothing. Split it until each theme points to a team or a fix.
- 03
Ignoring the “Other” group
When answers with no theme grow, a new topic is forming. Read them and add a theme when one keeps coming up.
- 04
Counting without quotes
A count shows size but not what people mean. Keep two or three answers beside each theme.
Frequently asked questions
A code is a short tag for one idea in an answer, such as “tracking link broken”. A theme groups related codes under one topic, such as delivery, and is what you count and report.
Yes. A model can file answers into themes quickly and in any language. It works best with a clear list of themes and a description of each, and with a person who checks a sample and corrects mistakes.
Enough that each theme leads to a decision, and few enough to read on one screen. Group them under broader categories when the list grows.
No. Usertick files answers into your organization's list of categories and themes. On every plan you can add a theme, up to the plan's limit, and the draft shows how many recent answers it would catch before you save it.
Thematic analysis says what an answer is about, and sentiment analysis says how it feels. Together they show which topic is making customers unhappy and whether it is getting better.
Listen better, act faster, and build what truly matters.
Start free, see what your customers mean, and get each answer to the person who can act on it.
- 150 responses a month on Free
- Every channel and AI analysis
- No card needed