Anyone whose real operational record is a Google Sheet with six tabs, a tab everybody knows not to touch, and a column of dates that are secretly text.
Pull Google Sheets into RowFold tables you own — filter them, report on them, link them to everything else in the workspace.
And it stays current: importing again updates the same tables, adding new records and updating changed ones, so links from your other tables hold. Keep in sync does that for you hourly or daily.
What it doesn't do: There are no automation steps that write back to Google Sheets. It has no say in what a booking page offers. An import only reads: nothing is written back to Google Sheets, a record deleted there stays here, and a sync runs hourly at most rather than the moment something changes. Anything else you have read about Google Sheets and RowFold, this page is the authority: it is generated from the connector's own code.
One tab becomes one table and the first row becomes the field names, so ten minutes tidying the sheet in Google saves an hour of renaming afterwards. The import is one way and it never writes back, which means the sheet is safe while you try this.
You'll know it worked: One new table per tab appears in the left rail. Tabs that are charts rather than grids, and empty tabs, are skipped and the import says so. Sort a date column: until you have set it to Date it will sort alphabetically, which is expected and is the reminder to set the type.
This is the part that makes the move worth doing. In a sheet, Orders and Customers are joined by a name typed the same way in both places and a VLOOKUP that breaks when somebody sorts a range. Here they can be joined properly, once, and the join then holds by itself.
You'll know it worked: Open a customer record and look for the linked orders on it. If the list is empty for a customer you know has orders, the matching column had a spelling variant, which is exactly the problem the relation is there to end.
Tabs pretend to be tables and cannot behave like them. The Orders tab holds a customer name typed slightly differently in four rows, and nothing in the sheet can say that those four are the same customer, because a sheet has no idea what a customer is.
Each tab arrives as its own table, and from there you can link them for real. One customer record, four orders pointing at it, so a rename fixes every order at once and totals stop depending on spelling.
A spreadsheet cannot stop anybody doing anything. One person sorts a range without the column beside it, someone else pastes over a formula, and the version that was right existed for about an hour on Tuesday.
A field has a type and rules, and a record is a row that cannot be knocked out of alignment by a sort. Every change is attributed and reversible, so the question is no longer who broke it but simply when it changed.
Sharing a sheet is all or nothing at a level anybody notices. Hiding a column is a suggestion, protected ranges are brittle, and the usual answer is a second sheet maintained by copy and paste for the people who should not see the margins.
Permissions are per table and per field, so the same data serves everybody at the level they should see it. Nobody maintains a redacted copy, because there is nothing to copy.
Everything downstream is manual. A new row means somebody notices and does the next thing — send the email, tell the warehouse, chase the approval — and the process only exists in the head of whoever usually does it.
Rows are records, so they can trigger things. The chase, the notification and the status change run themselves, in a log you can read afterwards.
Because RowFold deliberately asked for the narrowest access Google offers: permission to open files you explicitly hand over, rather than permission to see your Drive. That is a meaningfully smaller thing to grant, and the picker is how a file gets handed over. The trade is real — a pasted link to a file you never picked will not open, by design — and it is the right way round for a tool you are about to put your operational data into.
The values come across as text, not the formulas. That is usually the right outcome, because most spreadsheet formulas exist to fake things a database does properly: a lookup becomes a real link, a running total becomes a rollup that cannot go stale, and an IF chain becomes a formula field every row shares instead of one that was dragged down 400 rows and stopped at 380.
It can, one way. Importing the spreadsheet again updates the tables it made, and Keep in sync does that hourly or daily. A sheet has no row IDs, so each row is matched by the value in its first column: change that cell and the row arrives as a new record, with the old one left where it was. Nothing writes back to Google, and an edit made here to a column the sheet fills is replaced on the next run, so the cleanest shape is still that the table becomes the original and the sheet stops being maintained.
The first row is read as the field names, so a sheet with a title banner or a blank line above the headers will produce fields called things like Column B. It is worth a minute in Google first: delete the decoration above the headers, and make sure each column actually has a name.
Up to 10,000 rows and 200 columns per tab. Past either limit the import stops at the cap and tells you so in its warnings, rather than quietly leaving the rest behind. For most operational sheets that is plenty; a tab genuinely larger than that is usually a log that wants splitting by year, or a sign the data should arrive another way, such as the API or a CSV import.
Yes, and that is the part worth doing properly. Bring the tabs across, then add a relation between the tables — Orders to Customers, Line Items to Orders — and the joins your sheet was faking with VLOOKUP become real. Do it once and every later report gets it for free.
There is no template to wait for. Type a sentence about what you run and RowFold assembles the tables, links, views and dashboard around it — then you change whatever does not fit.