Quick answer
Quantitative data is information recorded as numbers that count or measure something, such as visitors, orders, load time or order value. A number needs a definition and a denominator before it can be compared, and discrete counts and continuous measures need different summaries.
Formula: Mean = sum of values ÷ number of values; median = middle value of the sorted list
What is quantitative data?
Quantitative data is information recorded as numbers that count or measure something, so that arithmetic on it makes sense: 1,240 visitors, 3 items in a cart, 2.4 seconds to load a page, $59.90 per order. It answers questions of "how many", "how much" and "how often", and it is what dashboards, reports and statistical tests are built on.
Two conditions separate quantitative data from numbers that are really labels. First, the number must be an amount: a ZIP code or an HTTP status code is written in digits but is a categorical variable. Second, the number needs a definition and a denominator before it can be compared with anything. "Average session duration: 2 minutes" says little until you know which sessions were counted and how duration was measured.
Quantitative vs qualitative data in one paragraph
Qualitative data describes qualities in words, images or categories: interview notes, support transcripts, the reason a customer gave for cancelling. Quantitative data measures in numbers. The two answer different questions: numbers tell you that checkout completion fell from 62% to 48%; qualitative data such as session notes or chat transcripts tells you why. The full comparison, with research methods for each, is in qualitative vs quantitative data.
Discrete vs continuous data
Quantitative data comes in two forms, and the form decides how you summarize it:
| Discrete data | Continuous data | |
|---|---|---|
| What it is | Counts of whole units | Measurements on a scale |
| Possible values | 0, 1, 2, 3… (no values between) | Any value in a range (2.41 s, 2.413 s…) |
| Website and app examples | Page views, orders, sessions, items in cart, crashes | Session duration, page load time, order value, scroll depth |
| How it is collected | Counting events | Measuring time, size or money |
| Natural summaries | Totals, rates per user or per session, frequency distribution | Median, percentiles (P75, P90), mean with spread, histogram |
| Typical trap | Counting per session when the question is per person | A mean pulled up by a few extreme values |
Money is technically discrete (it moves in cents) but is treated as continuous because the steps are tiny. Quantitative data can also be classified by level of measurement: interval data has equal steps but no true zero (temperature in °C), ratio data has a true zero so ratios make sense (revenue, duration). Almost all product metrics are ratio data. Ordered ratings such as a 1–5 score are not quantitative in this sense; see ordinal variable. The wider map of types is in data types.
10 examples of quantitative data on websites and apps
| Metric | Discrete or continuous | Unit | Denominator to state | Summarize with |
|---|---|---|---|---|
| Visitors | Discrete | People | Time range; humans only or including bots | Total, trend |
| Page views | Discrete | Views | Per visitor or per session | Total, per-session average |
| Orders | Discrete | Orders | Per visitor (conversion rate) | Total, rate |
| Items in cart | Discrete | Items | Per order | Distribution, mean |
| Crashes | Discrete | Events | Per active user or per session | Crash-free share |
| Session duration | Continuous | Seconds | Which sessions are included | Median, P75 |
| Page load time | Continuous | Milliseconds | Page views measured | P75, P90 |
| Order value | Continuous | Currency | Per order | Median, mean, sum |
| Scroll depth | Continuous | Percent of page | Page views of that page | Distribution |
| Active days | Discrete | Days | Per person over a window | Distribution, median |
Every row has a denominator column because the same count gives different answers with different denominators. 3,000 page views from 1,000 visitors is 3.0 per visitor; from the same visitors' 1,500 sessions it is 2.0 per session. Both are correct; they answer different questions.
How quantitative data is collected
- Event tracking. A script or SDK sends an event each time something happens (page view, click, purchase) with numeric properties such as price or duration. It gives complete counts for instrumented actions, and nothing for actions nobody instrumented.
- Server and database records. Orders, payments and subscriptions in your own systems. Usually the most reliable source for money, but disconnected from behavior unless the user IDs match.
- Server logs. Every request the server received, including bots. Good for crawl and error analysis; noisy for counting people.
- Surveys with numeric answers. Age, number of employees, hours per week. Self-reported, so subject to rounding and recall errors; rating scales collected this way are ordinal, not quantitative.
- Experiments. Assign users to variants and measure an outcome. The design, not the data type, is what makes a causal claim possible.
- Public and third-party datasets. Census figures, industry benchmarks. Check their definitions before comparing with your own numbers.
Mean vs median on the same data: a worked example
Sample data: nine sessions on a landing page, durations in seconds, sorted:
12, 25, 31, 40, 48, 55, 70, 95, 1,224
- Mean: (12 + 25 + 31 + 40 + 48 + 55 + 70 + 95 + 1,224) ÷ 9 = 1,600 ÷ 9 = 177.8 seconds.
- Median: the middle (5th) value = 48 seconds.
One session, likely a tab left open, makes the mean 3.7 times the median. Remove it and the mean of the other eight drops to 376 ÷ 8 = 47.0 seconds, while the median barely moves (from 48 to 44). Continuous data about human behavior is usually right-skewed like this, so the median and percentiles describe the typical visitor better than the mean. The mean vs median guide goes deeper on when each is right.
The denominator matters just as much. Sample data: 2,000 sessions with 224,000 seconds of total recorded time. Team A reports average time per session using only the 1,400 sessions that viewed two or more pages: 224,000 ÷ 1,400 = 160 seconds (2:40). Team B divides by all 2,000 sessions: 224,000 ÷ 2,000 = 112 seconds (1:52). Same data, same arithmetic, a 48-second disagreement, purely from the definition. How analytics tools measure duration, and why single-page sessions are tricky, is covered in average session duration.
How to summarize quantitative data correctly
- Write the definition next to the number. "Median session duration, sessions with 2+ page views, humans only, last 28 days" can be compared; "time on site" cannot.
- Match the summary to the form. Totals and rates for discrete counts; median and percentiles for skewed continuous measures; the mean when you need totals to reconcile (revenue per order × orders = revenue).
- Keep the count beside every rate. A 12% conversion rate from 25 visitors and from 25,000 visitors are different facts.
- Look at the distribution once. A histogram shows skew, outliers and odd spikes (such as many zeros from a tracking gap) that no single number reveals.
- Use the same denominator over time. If the definition changes, annotate the date and do not compare across it.
Common mistakes with quantitative data
- Letting a long tail drive the average. One extreme session or one enterprise order can move a mean more than thousands of typical values.
- Mixing denominators. Per session vs per person vs per page view; "all visits" vs "engaged visits".
- Counting bots as people. Crawler traffic inflates visitors and page views and deflates conversion rates.
- Averaging codes or ratings. Numbers that label categories are not quantitative; averages of 1–5 ratings need care.
- Confusing percent with percentage points. A rate going from 4% to 5% is a 1-point increase and a 25% relative increase.
- Comparing sums over unequal periods. February has fewer days than March; compare daily averages or same-length windows.
Measuring quantitative data in TapCub
The session-duration example shows that the aggregation is part of the question. In TapCub the aggregation is chosen explicitly, so the same events can answer several questions:
- Pick the aggregation. In All-platform analytics › Behavior › Metrics & SQL, a metric is an event plus an aggregation: total, uniques, average per subject, active percent, sessions, first time, or a property's sum, average, median, 90th percentile, distinct count, min or max. For a
purchaseevent with anorder_valueproperty, sum gives revenue, average and median give two answers to "typical order", and the 90th percentile shows how large big orders get. - Fix the definition once. Save the metric with its own filters, and give it a format (number, percent, duration or currency). Formula metrics combine saved metrics, such as buyers divided by active people, so the denominator is written into the definition rather than retyped per chart.
- Check the shape. All-platform analytics › Behavior › Behavior analysis has a distribution view: pick a numeric property, session duration or pages per session, and the histogram shows each bucket's count and share with the mean and the median side by side, which is the comparison in the worked example.
- Keep bots out of the denominator. Recognized bots and AI crawlers are counted separately under Web analytics › Bots, so they do not inflate visitor counts or deflate conversion rates.
See also: Quantitative Research Examples: Question, Sample and Test.
See also: What is discrete data? Examples and how it differs from continuous.
Sources
- NIST/SEMATECH e-Handbook of Statistical Methods, 1.3.5.1 Measures of Location (mean, median and their behavior on skewed data).
- NIST/SEMATECH e-Handbook of Statistical Methods, 1.3.3.14 Histogram.
- 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: Quantitative 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.