Docs · guide

How to create a pivot table in Google Sheets, and what it does alone

By Alberto Gulotta · Updated 4 September 2026 · 9 min read

Making one is short: select the data, Insert › Pivot table, then add fields to Rows, Columns and Values in the side panel. Google states one requirement above it, marked Important — “Each column needs a header.” That is the whole of the entry price.

It is worth pausing on how short that requirement is. Microsoft publishes five conditions for the same job, and the Excel guide is largely about getting a sheet into their shape before anything else. Google asks for headers. The two programs are not describing different data — they are documenting different amounts of it, and if your sheet came from somewhere else you will still meet every problem the Excel guide lists, just without being warned about them first.

This is the difference that changes how the thing is used. In Excel a pivot table is something you build, then maintain: add rows to the source and the summary is quietly out of date until somebody refreshes it, and there is no mark on the page saying so. In Sheets the summary follows the data. A sheet left open on a screen keeps telling the truth, and a stale total is not one of the ways this can go wrong.

Changing one afterwards is the same panel: the Edit button under the table reopens it, fields are moved by dragging them between categories, Remove takes one out, Select data range points the whole thing somewhere else, and Clear all empties it without deleting the sheet. Nothing here is destructive to the source, which is the reason a pivot table is the safe way to look at a spreadsheet somebody sent you: it answers questions about the data without touching it.

Not covered here. It will not tell you the two programs behave the same. One refreshes itself and the other has to be told to, and both companies say so in writing.

It will not pretend Google’s single requirement means the data can be any shape. It means Google documents less, not that a spreadsheet built to be read will summarise cleanly.

And it will not leave the automatic suggestions unmentioned. A table you did not choose is still a table somebody will read as your analysis. What holds instead is simple: the header requirement, the automatic insertion of high-confidence suggestions and the switch that disables them, the self-refresh behaviour, the calculated-field procedure with its quoting rule, and the double-click route to a cell’s source rows are quoted from Google’s own Docs Editors help, and the refresh requirement from Microsoft’s, listed below with the dates they were read.