Qualitative vs Quantitative Data: What Each Tells You, and How to Use Both

Quantitative data counts and measures; qualitative data explains why. This guide compares the two, lists examples of each in product and website work, and walks through an onboarding problem where the numbers find the drop and interviews and replays explain it.

The difference

Quantitative data is anything you can count or measure: 1,200 sign-ups, a 55% completion rate, 38 seconds on page, a satisfaction score of 7.2. It answers how many, how often and how much, and it can be added up, averaged and compared.

Qualitative data is descriptive: an interview transcript, a support conversation, an answer to "what almost stopped you from signing up?", a recording of someone hunting for a button. It answers why, how and what did they expect, and it is read and interpreted rather than added up.

In product and website work you almost always need both. Analytics tells you that 45% of new accounts never connect a data source. It cannot tell you that most of them do not have permission to create an API key.

Side by side

The table compares the two on the points that decide which to use. A useful rule: use quantitative data to find where to look and to measure whether a change worked; use qualitative data to understand what is going on and to come up with the change.

Use quantitative to choose where to look, qualitative to understand it.

Examples of each in product and website work

Quantitative: pageviews, unique visitors, conversion and bounce rates, funnel step completion, retention by cohort, event counts, revenue per user, survey ratings (a 1–10 score is a number), load times, rage-click counts.

Qualitative: user interviews, usability test notes, open-ended survey answers, support tickets and chat conversations, sales call notes, app store reviews, session replays, the comments field on a cancellation form.

Some sources are both. A heatmap is a count of clicks, but you read it like a picture. A cancellation survey has a multiple-choice reason (quantitative) and a free-text box (qualitative).

A worked example: why new accounts stall

A B2B tool sees 1,200 sign-ups in a month. The funnel shows 660 of them connect a data source within seven days, a 55% completion rate. That is the quantitative finding: the step loses 540 accounts, and it is the largest drop in onboarding.

The team then watches 25 session replays of accounts that reached the step and left, and runs 15 short interviews. The notes are coded: each session or interview gets one main reason. The chart shows the result. Sixteen of the 40 (40%) needed someone else, usually an admin, to create the key; ten could not find where the key was; eight were evaluating and had no real data to connect yet.

Coding turned qualitative notes into counts the team could prioritise. The fix for the top reason was an "invite a teammate to finish setup" link on the step, which a numbers-only view would never have suggested. The quantitative metric (step completion) then measures whether it worked.

How to combine them

1. Start with numbers to pick the question. Find the step, page or segment where the gap is largest.
2. Go qualitative on that spot only. Watch replays, read conversations and interview people from that step. Five to ten interviews often show the main reasons.
3. Code the answers. Give each note one main reason, count them, and check whether the top reasons cover most cases.
4. Change one thing and measure it with numbers. Compare the step completion rate before and after, ideally with a test.

FAQ

Is survey data qualitative or quantitative?

Both. Ratings and multiple-choice answers are quantitative; free-text answers are qualitative. Ask one rating and one open question, and code the open answers.

How much qualitative data is enough?

For finding the main reasons behind one problem, 5–15 interviews or 20–40 replays are usually enough to see the same reasons repeat. For measuring how common each reason is across all users, follow up with a survey.

Can qualitative data be analysed with numbers?

Yes, by coding: tag each item with a category, then count categories. Keep the original quotes next to the counts; they explain the reason better than the label.

Is session replay qualitative data?

A replay of one session is qualitative: you watch and interpret it. Counts across replays, such as how many had rage clicks or errors, are quantitative.

Related: UX analytics and session replay.

See also: What is qualitative data? Types, examples and how to analyze it.