Column statistics summary
Upload your sheet and download a report with one row per column: type, total, blanks, distinct values and, for numeric columns, min, max, sum, mean and median.
What it is for
Before analysing a spreadsheet you don't know, it pays to look at it from above:
how many rows each column has, how many are empty, how many distinct values
there are and, for the numeric ones, roughly where the numbers sit. It is the
classic describe, the first glance that stops you building an analysis on data
that isn't what it looks like.
In Excel that means a battery of COUNTA, COUNTIF, AVERAGE, MEDIAN and
MAX, repeated column by column. Here you upload the file and download the table
already built.
What each row of the report tells you
The result is an .xlsx with one row per column of your sheet and these
measures:
- Type — Numeric, Date, Text or Mixed. "Mixed" is a red flag: it means the column blends numbers with text, and that is nearly always a data problem worth checking.
- Total — how many data rows there are.
- Blanks — how many of those rows are empty in this column.
- Distinct — how many different values it holds. Handy for telling whether a column is a category (few distinct) or an identifier (all distinct).
For numeric columns it also adds min, max, sum, mean and median.
Why mean AND median
Both are given on purpose, because they tell you different things.
The mean is the ordinary average, but it is sensitive to extreme values: a single giant order or a typo can drag it and hand you a figure that represents almost no row. The median is the middle value once everything is sorted: half the data sits below it and half above, so one extreme does not move it.
When mean and median are close, the data is reasonably symmetric and the mean is reliable. When they drift far apart, extreme values are pulling on the mean and the median is the more honest figure. Seeing them together tells you, at a glance, whether you can trust the average.
Before you trust the numbers
This summary describes the data as it arrives. If a column that should be numeric shows up as Text or Mixed, it has numbers stored as text and the numeric measures will skip it. Run it through convert text to number first and regenerate the summary.
If what you want is to count rows by category — how many orders per country, how much each rep bills — the right tool is not this one but group and count. And if you suspect the sheet arrives messy, start with the data quality report.
Frequently asked questions
What counts as a "numeric" column?
Only those holding numbers (or booleans) and no text or dates. A single text value tips the column into "Mixed" and it loses the numeric measures: that is how it warns you the column isn't clean.
Do empty cells affect the mean?
No. Mean, median and sum are computed only over cells that hold a number. Empty cells are counted separately, in the "Blanks" column, so you can see how many there are.
Does it count distinct values ignoring case and spaces?
Distinct values are counted exactly as they appear, without normalising. If you
suspect Madrid and madrid are being counted as two, clean the column first
with clean data.
Does it change my file?
No. Your sheet is left untouched. What you download is a new report with the summary; the original stays intact.
Does it work with CSV and legacy .xls?
Yes. The report always comes out as .xlsx, whatever format you upload.