You open a workbook and get:

We found a problem with some content in ‘report.xlsx’. Do you want us to try to recover as much as we can?

Click Yes, and Excel “recovers” the file — usually by deleting something. Click No, and nothing opens. Either way you do not learn what was wrong.

This message is different from the generic “file is corrupt” dialog. It means the package is structurally valid and one part failed validation. That is a much better situation, and it is diagnosable.

What actually triggers it

The four causes, in rough order of frequency:

1. A broken external link or defined name. Workbooks that reference another file by path, or that carry hundreds of stale defined names from deleted sheets. Excel validates these at load and fails the whole workbook if one is malformed.

2. Damaged sharedStrings.xml. Every text value in the workbook is stored once in a shared string table. A single broken entry invalidates the table and the error follows.

3. A stale pivot cache. The pivot’s cached source data no longer matches the underlying range. Common in workbooks edited on several machines or after a column was deleted.

4. Invalid characters in a cell or sheet name. Usually the result of pasting from a web page or another system that emitted control characters.

Getting the actual error detail

Excel writes the specifics to a log instead of the dialog:

  1. Open the workbook and click Yes to recover.
  2. File → Info → (bottom), look for Show Repaired Records or check %TEMP%\error...xml.

The log lists the failing part, for example:

Repaired Records: Worksheet properties from /xl/worksheets/sheet2.xml
Removed Records: DefinedName from /xl/workbook.xml

That line is the answer. sheet2.xml means look at that sheet; DefinedName means the name manager, not the data.

Fixing each cause

Broken links and defined names. Formulas → Name Manager. Sort by the Refers To column, delete anything showing #REF!. Then Data → Edit Links → Break Link for any dead external reference. Save and reopen — this alone fixes a large share of cases.

Damaged shared strings. This is the one that needs surgery:

  1. Copy the file, rename the copy to .zip, extract.
  2. Check xl/sharedStrings.xml — look for a truncated final entry or stray characters.
  3. Rebuild the ZIP from the extracted contents (start with [Content_Types].xml), rename back to .xlsx.

If the table is badly damaged, an engineer can rebuild it from the individual sheet XML, because each cell still carries the index into the table.

Stale pivot cache. Select the pivot table → Analyze → Refresh if it opens at all. If it does not, delete the sheet in a copy of the file and re-create the pivot from the source range.

Invalid characters. If the file opens in “recovered” mode, sort or search for cells with odd characters and clean them. A quick way to find them: =LEN(A1)=LEN(TRIM(CLEAN(A1))) returns FALSE for rows with control characters.

The diagnostic everybody misses

Rename to .zip and open xl/workbook.xml. The <sheets> block lists every sheet and its relationship ID. If Excel’s repair log mentioned a sheet you did not know existed — hidden, very hidden, or left over from a deleted chart sheet — that is your problem, and deleting that one sheet usually ends the error.

When the fix is a rebuild

If two or more of the above apply, the workbook is better rebuilt than patched. Keep the data (values, formulas), rebuild the structure, and re-create pivots and charts from scratch. It sounds slow; it is usually faster than fighting the error a third time.

FAQ

Is it safe to click Yes and let Excel recover the file?

Safe for the file, lossy for the contents — Excel deletes whatever failed validation. That may be one stale defined name (no loss) or an entire sheet’s worth of formatting. Save the recovered copy under a new name so you can compare it against the original before you rely on it.

The error comes back after I save the repaired file.

Then the cause was not removed, only the symptom — the same trap as a PowerPoint deck that re-repairs itself. The usual suspects are a defined name that is re-created by an add-in on open, or an external link that Excel re-adds on save. Check Formulas → Name Manager immediately after opening the repaired file: if the #REF! entries are back, the fix has to happen in the add-in or in the source workbook, not in this one.

Can this error lose actual cell data?

Rarely. Validation failures hit metadata, caches and references far more often than cell values. The one exception is a damaged sharedStrings.xml, where text values live — that is the case worth handing to someone who can rebuild the table from the sheet XML rather than accept the loss.

If the repair log names a part you cannot identify, send the workbook over — reading that log is a two-minute job with the right tooling, and it usually saves the rebuild entirely.

Need the workbook as it was, formulas and all? Send the original file — before any repair attempt — to help@easyofficefix.com. You get the repair log interpreted, a flat price, and a straight answer on whether a rebuild is the better call.