Spreadsheets leak in more places than any other format
A workbook is the format most likely to carry personal data somewhere you are not looking. Deleting the visible column addresses one copy of the value and leaves several others:
- Hidden rows, columns and sheets. Hidden is a display attribute. The data is present in the file and one right-click away.
- Formulas. A concatenation or lookup can spell out an email address in the formula text even when the displayed result is something else.
- Defined names and data-validation lists. Dropdown source ranges frequently sit on a sheet nobody opens.
- Comments and notes. Attached to cells, printed nowhere by default.
- Workbook properties. Author, company, last-modified-by.
- Charts, embedded images and pivot caches. A pivot table keeps its own snapshot of the source data, which survives the deletion of the source itself.
This is why "I removed the personal columns before sending it" so often turns out not to be true.
What DokMine returns
A valid .xlsx. Formatting, formulas, sheet structure and number formats
are preserved — the workbook still opens and still calculates. Matched
values are blacked out or replaced with a consistent typed token, so a
column of customer emails becomes [EMAIL_1] through
[EMAIL_n], with the same address always mapping to the same
token. That keeps the file usable for analysis: you can still count
distinct customers, you just cannot identify them.
Where sheet-level changes are needed, references are rewritten before
they are applied, so formulas pointing at a renamed sheet do not break
into #REF! errors.
How to do it
- Upload the .xlsx — or a batch, if you are cleaning a whole export directory.
- Choose the detectors: emails, phone numbers, person names, URLs, card numbers, IBANs, IP addresses.
- Extract first. On a spreadsheet this is worth doing every time — the per-type counts tell you immediately whether something is hiding on a sheet you had not considered.
- Redact, download, then open the workbook and check the sheets, the formulas and the pivot tables.
Limits worth knowing
Free-text columns are harder than structured ones: a note field with a name buried mid-sentence depends on name detection, which is never reliable. Very large workbooks may hit the per-document extraction cap or the file-size ceiling on your plan — see plans and quotas. Values assembled at calculation time from several cells may be matched in one place and not another.
DokMine makes no guarantee of complete detection or removal. Open the returned workbook and review it — including hidden sheets — before it leaves your machine.