Convert ODS to XLSX

An ODS is a complete, standards-published spreadsheet and there is nothing wrong with it. What refuses it is everything downstream: the accountant whose practice software imports Excel workbooks, the payroll bureau’s template, the funder’s budget form, the ERP import screen whose file picker lists one extension, the client who has never installed anything other than Office.

None of those will change for one file, and the argument is not worth having when the fix takes seconds. Converting outward is a courtesy to the recipient and an insurance policy for you: an email that says the attachment would not open costs more time than the conversion, and it always arrives after the deadline rather than before it. There is also a quieter cost to sending an unfamiliar format, which is that some recipients will not say anything at all. They will file the attachment, fail to open it, and the submission is never processed at all.

Calc can save directly as .xlsx and it is the same engine that runs here, so when the program is already open the fastest route is Save As. Saying so costs this page a conversion and is plainly true.

The times it is not the fastest route are specific and common. The machine is locked down and you cannot install anything. The file arrived on a phone. You are on somebody else’s laptop. Or there are forty spreadsheets and opening each one, waiting, choosing a format and dismissing the keep-format dialogue is genuinely an afternoon. Those are what this page is for. It is the same engine either way — the conversion here runs LibreOffice on a server, which is precisely the program you would have used on your own machine, so the output is the file Save As would have produced.

An ordinary workbook. Sheet names and order are preserved, cell formatting comes across, column widths and row heights are kept, frozen panes and print ranges survive, and named ranges continue to resolve. The file is written by a full office engine, so Excel opens it without a repair prompt and without the yellow banner that makes people distrust an attachment.

That last point is worth more than it looks. A recipient who sees a warning the first time they open your file will remember the warning and not the spreadsheet, and half of them will reply asking whether it is safe rather than reading it. A clean open is the difference between a document and an incident. If the workbook is going to somebody in a regulated environment — an auditor, a bank, a funder — that matters more still, because their attachment policy is not theirs to relax and a file that triggers a prompt may be deleted by a gateway before they see it.

The conversion rewrites each expression from the OpenDocument formula language into the syntax Excel expects, and the results stay live: the recipient can change an input and watch the totals move, which is usually the whole reason they wanted a spreadsheet rather than a PDF.

The common ground is very wide. Arithmetic, aggregation, lookups, conditionals, text handling and date functions all have direct counterparts and cross without incident. Array formulas and anything referring to another workbook are the two areas worth opening the result to inspect, because those depend on context that does not travel with the file. A link to a spreadsheet sitting on your own drive is the classic case: it resolves perfectly on your machine, points at nothing on theirs, and the recipient sees an error where a figure should be with no way to know what it was.

LibreOffice ships functions that are not in the OpenDocument formula standard and not in Excel. When such a formula is exported, it is written under a namespaced name that Excel cannot resolve, and the cell shows a name error rather than a plausible wrong number.

That is the good failure mode, and it is worth appreciating: an error is visible, a silently different value is not. Before sending a model outward, open the converted file and look for error cells. If any turn up, replace the offending formula with an equivalent built from standard functions, or paste that column as values if the recipient only needs the answer. The second option is underrated for anything being submitted rather than collaborated on. A funder checking a budget wants the numbers to be right and reproducible, not the model to be live, and a sheet of values cannot break on their machine at all.

A number is stored as a number in both formats, so nothing about the underlying values changes. What changes is presentation: a currency symbol, a thousands separator, a decimal comma or point, and the order of day and month in a date are all resolved by the settings on the machine that opens the file.

This is the one thing that reliably confuses people when a spreadsheet crosses a border. If the recipient is in another country, write dates in a format that is unambiguous in every locale, and label a currency column in its header rather than relying on the symbol. That advice is older than either file format and it still prevents the most expensive kind of misreading. The one that actually costs money is a decimal separator: a figure entered as 1.500 meaning one and a half thousand, read by a machine configured the other way as one and a half. Nothing in the file is wrong and nothing warns anybody.

Charts carry their type, series and data references across and pick up Excel’s styling once they are there, so a bar chart is still a bar chart of the same numbers and may not be the same shade of blue. Pivot tables generally rebuild against their source range. Conditional formatting survives for the ordinary comparison rules.

What does not survive is the elaborate end: the more unusual chart types, some icon-set rules, and anything that depended on a Calc-specific feature with no Excel counterpart. If the spreadsheet is being sent to be looked at rather than recalculated, consider whether a PDF of the chart alongside the workbook would save an exchange of emails.

The pattern here is a submission: twelve monthly returns, a set of departmental budgets, a year of stock counts, all due at once in a format the portal names. Drop the folder rather than the files individually; each converts in turn, keeps its name with the extension changed, and the whole set comes back as one ZIP.

Check the one with the most formulas rather than the first one in the list, and check it in Excel if you have access to Excel. Verifying a file in the program that produced it tells you almost nothing about how the destination will read it, and the destination is the only opinion that counts here. If Excel is genuinely unavailable, the next best check is the portal itself: most upload forms validate and report a problem before anything is committed, and a rejected test upload two days early is information rather than a failure.

This conversion runs on a server rather than in your browser. Rewriting formulas between two spreadsheet formats requires a spreadsheet program, and there is no browser-sized library that does it correctly, so the file is sent over an encrypted connection to a container running LibreOffice headless.

That container has no outbound internet access. Each job gets its own scratch directory and its own LibreOffice profile, and the directory is deleted the moment the job ends, whatever the outcome; anything still running after sixty seconds is killed rather than left to hang. The free limit is 25 MB per file.

If the recipient is going to read the figures rather than work with them, a PDF removes every question raised on this page at once: no locale, no formula errors, no version differences, and no chance of them changing a number and quoting it back to you.

If the recipient is a system rather than a person, a CSV of the relevant sheet is usually what its import routine actually wants, whatever the upload page says about Excel. Send an XLSX when a human being will genuinely open the file and use it as a spreadsheet — that is the case this conversion serves well.