Drop an .xlsx or .ods here
How to use
What it is and what it's for
Upload an .xlsx or .ods and recalculate all its formulas, comparing the result with what the file had saved — the last thing Excel or LibreOffice calculated, which isn't always up to date (a cell changes and nobody recalculates, or the file got saved mid-recalculation). It also detects real circular references (A depends on B which depends on A) by following the actual dependencies, not guessing from the text. All in your browser: the file isn't uploaded to any server.
Why it matters
A Panko study found at least one error in 88 % of the spreadsheets audited. Reinhart and Rogoff published a 2013 economic policy study with an Excel error that excluded five countries from the calculation. The UK lost track of 15,841 COVID cases in 2020 because of an old .xls file's row limit. None of those errors would have needed a powerful computer to catch: they needed recalculating and comparing.
The engine: what it recalculates and what it doesn't
The recalculation uses a real formula engine (not an approximation), with support for most Excel functions. Two of them, OFFSET and INDIRECT, are marked «unsupported» instead of faking a result: they're references that resolve dynamically (which cell they point to depends on the calculation itself) and don't fit well with how this engine walks the dependencies. If your sheet uses them a lot, those specific cells can't be verified — the rest of the sheet can.
How circular references are detected
It doesn't search for «A1» inside the text of other formulas (that misses aliases, named ranges or cross-sheet references). It genuinely recalculates, cell by cell, following each reference; if a cell asks for its own value while it's still being calculated, that's a real circular reference, and every cell that's part of it gets flagged, not just the first one detected.
