Some sheets stop being spreadsheets.
A spreadsheet is the best tool ever made for figuring out what you’re even doing — free-form, instant, no schema to agree on first. Then other people start depending on it, and a cell that was a guess becomes a number the business quotes. That sheet has become a database. It just hasn’t got the guardrails of one yet.
How to read this page
We made this page, so read it that way. Every Google limit on it is Google’s own published figure, read on 2 August 2026 — and there is no price comparison, because Sheets comes with a seat most teams already pay for.
Six signs it stopped being a spreadsheet.
Not one of these is about size. They’re all about the sheet having become something other people rely on.
You’ve outgrown it when…
- Someone has broken a formula by pasting, and nobody noticed for a week
- The same client’s name is spelled three ways, and VLOOKUP is how you join
- There’s a tab called Sheet1 (2) FINAL v3
- You’d like some people to see rows and not edit them
- Something else needs to read it automatically, on a schedule
- You want to know who changed that, and when
What a database does instead
- A field has a type, so a date can’t become text
- Records link to records — no lookup formula to break
- One table, many saved views, no copies
- Viewer, editor and owner roles per database
- A webhook fires the moment a record changes
- A seven-day trash, and undo
Two or three of these in a month, and the sheet is already doing a database’s job without a database’s guarantees.
Ten million cells is a real number.
Most teams never reach these. They’re here because the first one is usually met from an unexpected direction: cells, not rows — a forty-column sheet runs out at two hundred and fifty thousand rows.
Size
API
Shape
Connected Sheets, the usual answer for bigger data, has its own published ceilings: 200,000 rows for a pivot table and 500,000 rows or 5 million cells for an extract.
What we left to the spreadsheet.
Ad-hoc arithmetic on a blank canvas — by design. A spreadsheet lets you put a number anywhere and a formula beside it. We make you say what a column is first, because that is the whole point: it’s what stops a date becoming text and a client’s name existing five ways.
So the deep formula language and pivot tables stay where they belong. Plenty of teams keep a sheet pointed at their data for exactly that, and export here takes one click. What we built is the part underneath — the structure the numbers sit on.
Download as CSV. Drop it here. Ten minutes.
Bring the sheet over
File → Download → CSV, one tab at a time, and our importer takes it from there — it reads CSV and Excel, guesses each column’s type, and shows you the guesses before anything is created. The part worth your attention is deciding which columns became links to another table; that’s the whole difference, and it’s a five-minute job on a normal sheet.
And leave the same way
The whole database comes back out in one zip — every table as CSV, every record, every attachment, with a manifest. Which is to say: back to a spreadsheet, any day, free. If this turns out not to be the tool, you have lost an afternoon.
The door swings both ways. It’s the only honest way to ask you to walk through it.
Try one sheet. Keep the rest.
Move the tab that’s become load-bearing and see how it feels — every feature, free while we’re in beta.