Average Calculator

"Average" is three different statistics wearing one word. The calculator above reports all of them — mean, median and mode — side by side, because which one you quote changes the story the numbers tell.

How it works

Paste or type your numbers separated by commas, spaces or new lines; mixed separators are fine, so a column copied out of a spreadsheet works as-is. Anything that is not a finite number is silently dropped rather than treated as zero, so a stray label or a N/A cell will not drag your mean down. The tool then reports seven figures:

• Mean — the sum divided by the count.
• Median — the middle value once sorted, or the average of the two middle values when the count is even.
• Mode — the most frequent value. If several values tie for most frequent they are all listed; if every value is unique, the mode is reported as None rather than as a meaningless first item.
• Sum, Min, Max and Count — the raw aggregates, useful for sanity-checking that everything you meant to paste actually parsed.

Everything runs in your browser. Nothing is uploaded, which matters when the list you are pasting is salary data, response times or exam marks.

Mean, median or mode?

The mean is the only one of the three that uses every value, which is exactly why it is fragile: a single outlier moves it. Ten requests served in 50 ms and one that took 30 seconds give a mean of about 2.8 seconds — a number that describes none of the eleven requests. The median stays at 50 ms and is a far more honest summary.

The rule of thumb: use the mean when values are roughly symmetric and every observation should pull its weight (temperatures, test scores, measurement error). Use the median when the distribution has a long tail (incomes, latencies, file sizes, time-to-resolution). Use the mode for categorical or heavily rounded data, where "the most common answer" is the question you are actually asking.

This is not a pedantic distinction in engineering. Average response time is one of the most misleading metrics a team can put on a dashboard, because the users having a bad time are precisely the ones the mean erases. Percentiles exist for that reason, and the guide to performance testing explains how p95 and p99 are read and why a passing mean can hide a failing service. The broader picture of which numbers to instrument, and how they relate to user-perceived speed, is covered in application performance.

Things worth checking

Count first. Before you trust any of the statistics, check that Count matches the number of values you expected. A mismatch means something in your paste did not parse — most often a thousands separator, since 1,250 is read as two numbers, 1 and 250.

Mode is often absent, and that is correct. Continuous measurements almost never repeat exactly, so an empty mode on a list of timings is the expected answer, not a bug.

Precision. Figures are computed with standard double-precision arithmetic. Summing a long list of decimals accumulates a tiny rounding error in the last digits, so treat the trailing decimals of a mean over thousands of values as noise rather than signal.

An average of averages is not an average. If you already reduced each group to its mean, averaging those means only gives the right answer when the groups are the same size. Go back to the raw values and paste those instead.