Quick answer
Qualitative data is non-numerical information that describes qualities, opinions, reasons and experiences, recorded as words, images, audio or categories. It answers why and how, where quantitative data answers how many and how much.
What is qualitative data?
Qualitative data is information that describes qualities rather than measuring quantities. It is recorded as words, images, audio, video or categories: what a customer wrote in a chat, why a user abandoned a form, which plan a visitor chose. It answers why and how, where quantitative data answers how many and how much.
A funnel tells you that 46% of shoppers drop between checkout and payment. Qualitative data tells you they left because the shipping cost appeared only on the last step.
Qualitative vs quantitative data in one table
| Qualitative | Quantitative | |
|---|---|---|
| Form | Words, images, categories | Numbers, counts, measurements |
| Answers | Why? How? What is it like? | How many? How much? How often? |
| Typical sample | 5–30 interviews, hundreds of comments | Thousands of sessions or events |
| Analysis | Coding and themes | Statistics |
| Weakness | Hard to generalise; analyst bias | Shows what, not why |
The full comparison, with when to use each, is in qualitative vs quantitative data.
Types of qualitative data
By structure:
- Nominal (categorical) data. Labels with no order: country, browser, plan name, reason for cancelling.
- Ordinal data. Labels with a meaningful order but unequal gaps: "very unhappy" to "very happy", T-shirt sizes, priority levels.
- Binary data. Two categories: yes/no, subscribed/unsubscribed.
- Unstructured data. Free text, audio, video and images that have not yet been categorised. Most qualitative research starts here.
The nominal/ordinal split comes from S. S. Stevens' 1946 levels of measurement; ratings on an ordinal scale are often summarised with numbers, but the gaps between the labels are not equal, so treat averages with care.
Examples of qualitative data
- Interview and usability-test notes ("couldn't find the export button")
- Open-ended survey answers
- Live chat transcripts and support tickets
- App store reviews and social posts
- Session recordings and heatmaps of real visits
- Diary-study entries, photos, screenshots
- Sales call notes and lost-deal reasons
How qualitative data is collected
| Method | Typical size | Cost | Good for |
|---|---|---|---|
| In-depth interviews | 5–15 people | High | Motivations, mental models |
| Focus groups | 2–4 groups of 6–8 | High | Reactions to concepts and language |
| Usability tests | 5–8 users per round | Medium | Where and why people get stuck |
| Open-ended survey questions | 100s of answers | Low | Reasons at scale |
| Support chats and tickets | 100s per month | Low (already exists) | Real problems in customers' own words |
| Session replay | 10s of recordings | Low | What people actually did on the page |
Existing sources such as chat transcripts are underrated: nobody prompted the customer, so the language is unfiltered.
How to analyze qualitative data
The most common method is thematic analysis, described by Braun and Clarke (2006) in six phases:
- Familiarise yourself with the material: read everything once without coding.
- Generate initial codes: tag meaningful passages with short labels ("shipping cost surprise", "coupon rejected").
- Search for themes: group related codes.
- Review themes against the data: does each theme hold up, are two themes really one?
- Define and name themes so someone else could apply them.
- Write up: each theme with how often it appears and two or three representative quotes.
If two people code, have both code the same 20 items first and compare. Disagreements usually mean a code is badly defined.
A worked example: coding chat transcripts
Sample data: 120 pre-sales chat transcripts from an online shop in one month, coded by the main reason the visitor wrote in.
| Theme | Transcripts | Share | Representative quote |
|---|---|---|---|
| Shipping cost or delivery time | 34 | 28.3% | "Why is shipping only shown at the end?" |
| Size and fit | 27 | 22.5% | "Does the M run small?" |
| Payment failed | 19 | 15.8% | "Card declined twice" |
| Discount code not working | 15 | 12.5% | "WELCOME10 says expired" |
| Account and login | 11 | 9.2% | "Didn't get the reset email" |
| Other | 14 | 11.7% | |
| Total | 120 | 100.0% |
Two things happen next. The top theme becomes a hypothesis for the quantitative side: does the checkout-to-payment drop rise on orders where shipping is above $10? And the theme counts become a monthly metric: if "shipping cost" falls from 28% to 12% after showing the cost on the product page, the change worked.
Common pitfalls
- Cherry-picking quotes that support what you already believed. Count first, quote second.
- Leading questions in interviews and surveys ("How much did you like…").
- Treating 8 interviews as a percentage. "5 of 8 users" is a finding, "62.5% of users" implies precision you do not have.
- Codes that drift over the analysis. Write a one-line definition for every code.
- Ignoring the silent majority. People who complain in chat are not all customers; pair the themes with behavioural data.
Qualitative data in TapCub
TapCub does not run interviews or surveys. The qualitative sources it does provide sit next to the numbers:
- Live chat transcripts in Live chat › Inbox, with the visitor's pages and source in the side drawer, and hot pages in Live chat › Reports showing where chats start.
- Draft Q&A from past chats in the knowledge base, which groups visitor questions and agent answers from resolved conversations: a quick first pass at your common themes.
- Session replay (off by default, recorded for a sampled share of sessions once enabled), plus heatmaps and rage clicks, to see what people did where a funnel step drops.
A good routine: find the drop in a funnel, read 20 chats and 10 recordings from that step, code them, then check the biggest theme against the data.
Sources
- Braun, V. and Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77–101.
- Stevens, S. S. (1946). On the Theory of Scales of Measurement. Science, 103(2684), 677–680.
- Network of the National Library of Medicine, Data glossary: qualitative data.
Was this explanation clear?
We use your answers to fix and expand this page.
Couldn’t send. Please try again in a minute.
Thanks for the feedback. We’ll use it to improve this page.You’ve already sent feedback on this page. Thanks.
Product Analyst
Focuses on event analysis, conversion funnels, and retention, using clear metric definitions and structured analysis to explain product performance.