Excel to Google Sheets: What Actually Breaks
Moving a workbook from Excel to Google Sheets looks instant. What converts cleanly, what breaks silently, and the checks to run before you trust it.
Moving a workbook from Excel to Google Sheets looks instant — upload the file, it opens, done. Most of what actually breaks in that move does so silently, which is the expensive kind of broken: nobody notices until a total is already wrong in a shared sheet.
What actually happens when you move the file
Google Sheets opens an uploaded .xlsx file directly — there is no separate "convert" step to click. It will keep working on it in that uploaded format, or you can save it as a native Google Sheet through File > Save as Google Sheets, which is the version that picks up Sheets-specific features like Apps Script and live multi-editor collaboration.
What breaks, and does not announce it
- VBA macros do not come across at all. Google Sheets runs Apps Script instead, a different language entirely, so a macro-driven button survives the move visually and does nothing when clicked.
- A handful of Excel functions have no exact Sheets equivalent, or compute slightly differently — array-entered formulas and certain date functions are the most common source of a quietly different number after the move.
- Conditional formatting rules can shift or disappear outright on anything with more than one or two conditions.
- Links to other external workbooks break, since the file is no longer sitting on the same filesystem those links pointed to.
- Pivot tables sometimes need to be rebuilt from scratch rather than carrying across intact.
A worked example
Say a quarterly expense workbook has a SUMIFS total at the bottom and a macro button that resets the input form. After the move: the SUMIFS formula survives untouched, because Sheets supports SUMIFS natively. The macro button does nothing. And a different total two rows down, built on a date function that behaves slightly differently across the two platforms, is quietly off by a day’s worth of transactions — correct-looking, and wrong.
I moved this workbook from Excel to Google Sheets. Here is the formula from the old file: =NETWORKDAYS.INTL(B2,C2,1,Holidays). Does this behave identically in Sheets, and if not, show me the Sheets equivalent and explain exactly where the two differ.
An AI tool is genuinely good at this exact translation job — naming, function by function, which ones changed meaning across the two platforms. It cannot see your actual file unless you paste the real formula and the real error, which is why pasting the formula is worth more than describing the problem in general terms. How to Write a Prompt That Works on the First Try covers that habit directly, and OpenAI’s and Anthropic’s own prompting guidance both make the same point from the vendor side.
Where AI genuinely helps, and where it does not
Drafting the Apps Script replacement for a lost macro is a good use of it — Code for Excel covers the equivalent choice on the Excel side, between VBA, Office Scripts and Python, and the same logic of picking the right tool for the job applies once that code needs to live in Sheets instead. One real formula shown alongside the request gets a noticeably better draft than describing the macro’s purpose in the abstract.
What it is not good at is silently verifying that your converted totals still match the original. It will describe a translated formula as correct with total confidence, without ever re-running it against your actual numbers — a widely cited survey of hallucination in large language models covers why fluent and accurate are not the same property in a model’s output. A wrong total after a file move looks exactly as certain as a right one.
Checks before you trust the move
- Recompute at least one total by hand against the original Excel file, not the newly converted one.
- Compare row and column counts between the old file and the new one — a dropped row during upload is rare on a small sheet, less rare on a very large one.
- Open any formula that uses a date, array, or lookup function and confirm by hand that it returns the same answer it did before the move.
- Rebuild any macro-driven action as an Apps Script function, then run it once against a copy of the data before trusting it on anything live.
- Keep the original .xlsx file until every number you actually rely on has been checked once against it.
Deciding in advance what you would check, rather than discovering it after a number is already wrong in a shared sheet, is the same habit behind NIST’s AI Risk Management Framework — scaled down here to one converted workbook instead of a deployed system.
The same move-and-verify pattern shows up elsewhere in a spreadsheet’s life, not just a platform change — Excel Table covers what a structured table actually survives intact, and Excel Alternative is worth reading first if Google Sheets is a genuine switch rather than a one-off file you need to open for someone else. Online Excel Viewer covers the narrower case of just needing to see a workbook without editing it, and Can You Upload Excel to ChatGPT? covers the adjacent question of handing the file straight to a chat tool instead of Sheets. For turning a verified total into a sentence for a report once the move is done, Turning an Excel Formula Into Text is the next step.
The short version
The upload itself is the easy part and takes seconds. The move is only actually finished once you have recomputed a real total by hand, checked that no row went missing, and rebuilt anything that depended on a macro. Treat the first version that opens in Sheets as a draft, not a result, until those checks are done.
Coursium teaches that same habit — asking a precise question, and checking the answer before you trust it. Stay ahead of AI by learning the tools on your phone.