Quick answer
Descriptive statistics summarize a data set as it is: counts and frequencies, center (mean, median, mode), spread (range, standard deviation, interquartile range) and shape (skewness, percentiles), without generalizing beyond the data.
What are descriptive statistics?
Descriptive statistics summarize and describe the data you actually have: how many values there are, where the center is, how spread out they are, and what shape the distribution takes. They do not try to generalize beyond the data set; that is the job of inferential statistics.
Every analytics dashboard is mostly descriptive statistics: total sessions, average order value, median time on page, the share of visitors on mobile.
The four types of descriptive statistics
| Type | Question it answers | Common measures |
|---|---|---|
| Count and frequency | How many, and how often does each value occur? | Count, frequency table, percentage share |
| Central tendency | What is a typical value? | Mean, median, mode |
| Dispersion (spread) | How much do values vary? | Range, standard deviation, variance, interquartile range, coefficient of variation |
| Shape and position | Is the distribution symmetric? Where does a value rank? | Skewness, percentiles, quartiles, five-number summary |
For categorical data (device type, country, traffic source) you mostly use counts, percentages and the mode. For numeric data (sessions, revenue, load time) you use all four groups.
Worked example: 14 days of sessions
Daily sessions on a site over two weeks (sample data):
1,240 1,310 1,185 1,402 1,366 980 910
1,288 1,325 1,270 1,450 1,398 1,012 2,890| Statistic | All 14 days | Without the 2,890 spike |
|---|---|---|
| Mean | 1,359 | 1,241 |
| Median | 1,299 | 1,288 |
| Standard deviation | 471 | 173 |
| Coefficient of variation | 35% | 14% |
| Min / max | 910 / 2,890 | 910 / 1,450 |
| Q1 / Q3 | 1,199 / 1,390 | — |
| Skewness | 2.32 | — |
Reading it:
- One day dominates the mean and the standard deviation. The 2,890-session day (a newsletter send) lifts the mean by 118 sessions and nearly triples the standard deviation. The median moves by only 11.
- The five-number summary (min 910, Q1 1,199, median 1,299, Q3 1,390, max 2,890) shows the problem immediately: the max is far above Q3 while the min is close to Q1.
- Skewness of 2.3 says the distribution has a long right tail, so the median is a better "typical day" than the mean.
- The three lowest days are weekends. Descriptive statistics will not tell you why; a breakdown by day of week will.
Choosing the right summary
- Symmetric numeric data: mean and standard deviation.
- Skewed numeric data (revenue, time on page, load time): median plus interquartile range or percentiles such as P75 and P95. See mean vs median and percentile.
- Categorical data: counts, percentage shares and the mode. An "average device type" does not exist.
- Anything you will compare across periods: report the count alongside every rate, so readers can judge the base. A 50% conversion rate on 4 visitors is not a finding.
Always look at the shape before choosing. A histogram or box plot takes seconds and prevents most summary mistakes; Anscombe's quartet shows four data sets with identical means and variances that look completely different when plotted.
Descriptive vs inferential statistics
| Descriptive | Inferential | |
|---|---|---|
| Goal | Describe the data you have | Draw conclusions about a larger population or process |
| Output | Means, medians, counts, charts | Confidence intervals, p-values, tests, forecasts |
| Uncertainty | Not quantified | Quantified (margin of error, significance) |
| Example | "Average order value last month was $62" | "Average order value is between $57 and $67 with 95% confidence" |
Most analysis starts descriptive and becomes inferential when you ask whether a difference is real or will hold next month; see statistical significance.
Common mistakes
- Reporting the mean of skewed data alone. One large customer or one viral day changes the story.
- Averaging averages. The mean of daily conversion rates is not the period's conversion rate when daily traffic differs. Recompute from totals.
- Dropping the count. Percentages without denominators hide small, noisy segments.
- Treating outliers as errors by default. Check them first; see outliers in data.
In TapCub
Most TapCub reports are descriptive: counts, rates and averages by page, source, device or segment. The distribution report shows how a metric spreads across users in buckets, which is the quickest way to see skew before choosing between a mean and a median. Performance cards switch between average, median, P75 and P95. For anything the built-in reports do not compute, export raw events or use SQL; see types of data analytics for where descriptive analysis sits among the others.
Sources
- NIST/SEMATECH. e-Handbook of Statistical Methods: exploratory data analysis.
- OpenStax. Introductory Statistics, chapter 2: Descriptive Statistics.
- Tukey, J. W. (1977). Exploratory Data Analysis. Addison-Wesley (five-number summary and box plot).
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 Experience Specialist
Writes about product experience, practical workflows, and growth, turning complex processes into clear steps for everyday decisions.