Excel Damage Diagnostic

A long number was converted to scientific notation. Whether this is repairable depends on how many significant digits survived: "8.71E+12" describes a thirteen-digit number but only carries three of its digits, so the other ten are gone and no tool can invent them. This tool counts the surviving digits and says which case you are in.

Fix: If digits were lost, re-export from the source with the column formatted as text — repairing this copy is not possible.

Source: Microsoft Support — Keeping leading zeros and large numbers

Spreadsheets keep a maximum of 15 significant digits and replace everything after the fifteenth with zeros. A sixteen-digit-or-longer value ending in zeros has the exact shape that damage leaves, though a genuine number could end in zeros too — which is why this is reported as a signature rather than a verdict.

Fix: Check the value against your source of truth. If it was longer than 15 digits, the tail is gone and the column needs re-exporting as text.

Source: Microsoft Support — Keeping leading zeros and large numbers

Spreadsheets remove leading zeros so that values behave as numbers. The damage is only detectable in context: this value is exactly one digit shorter than the rest of its column, which is what a lost leading zero looks like.

Fix: Re-export with the column formatted as text. The missing zero can be restored here because the intended width is known from the column.

Source: Microsoft Support — Keeping leading zeros and large numbers

A product code that looked like a date was converted into one. "MAR1" becomes 1-Mar, "SEPT2" becomes 2-Sep. The classic case in the wild is gene names, but any SKU scheme that mixes three letters with digits hits it. The original text cannot be recovered with certainty because several codes map to the same date.

Fix: Re-export with the column formatted as text. The candidates listed here are possibilities, not an answer.

Source: Derived from the documented behaviour of spreadsheet number handling and from the shape the damage leaves behind in exported files. Each signature is matched against its own regression fixtures rather than assumed.

A long run of zeros or nines at the end of a decimal is the residue of binary floating-point arithmetic — 0.1 + 0.2 famously produces 0.30000000000000004. The intended value is almost always the rounded one, but the raw string will not compare equal to it.

Fix: Round to the precision the field actually needs, usually two decimal places for money.

Source: Derived from the documented behaviour of spreadsheet number handling and from the shape the damage leaves behind in exported files. Each signature is matched against its own regression fixtures rather than assumed.

A cell beginning with =, +, -, @, tab, carriage return or line feed is executed as a formula when the file is opened in a spreadsheet — and so are the full-width variants ＝ ＋ － ＠ under some locales. This is a security issue, not a formatting one: it can be used to exfiltrate the contents of open spreadsheets or to attack whoever opens the file.

Fix: Prefix the cell with a tab character inside the quoted field, or strip the leading character. Note that quote-based escaping can be undone when a spreadsheet re-saves the file.

Source: OWASP — CSV Injection

An error value such as #N/A or #REF! was exported as literal text. A formula failed somewhere upstream and the failure was shipped instead of the data.

Fix: Fix the formula at the source; there is no value here to recover.

Source: Derived from the documented behaviour of spreadsheet number handling and from the shape the damage leaves behind in exported files. Each signature is matched against its own regression fixtures rather than assumed.

Currency symbols, thousands separators, percent signs or accounting-style brackets for negatives are display formatting that has been written into the data. Feeds and APIs want the bare number.

Fix: Strip the formatting. Note that brackets mean a negative number in accounting format — (1,234.00) is -1234.00, not 1234.00.

Source: Derived from the documented behaviour of spreadsheet number handling and from the shape the damage leaves behind in exported files. Each signature is matched against its own regression fixtures rather than assumed.

A value became TRUE or FALSE. Product data that legitimately reads "true" is rare; more often a code, a flag column, or a two-letter country abbreviation was interpreted as a boolean.

Fix: Check the source column. If it was a code, re-export as text.

Source: Derived from the documented behaviour of spreadsheet number handling and from the shape the damage leaves behind in exported files. Each signature is matched against its own regression fixtures rather than assumed.

The value carries a non-breaking space, a zero-width character or stray surrounding whitespace. These survive copy-paste, break exact matching, and are invisible in every preview.

Fix: Trim and replace non-breaking spaces with ordinary ones.

Source: Derived from the documented behaviour of spreadsheet number handling and from the shape the damage leaves behind in exported files. Each signature is matched against its own regression fixtures rather than assumed.

None of the known spreadsheet damage signatures appear in this value.

Source: Derived from the documented behaviour of spreadsheet number handling and from the shape the damage leaves behind in exported files. Each signature is matched against its own regression fixtures rather than assumed.