Hold a long integer for review

The 15-digit boundary

Spreadsheet numbers hold about 15 significant digits. Beyond that, the last digits are rounded away when a value is stored as a number. To avoid producing such a number, the numeric normalizer does not convert an integer part longer than 15 digits. It leaves the cell's original text in place and records it under Values left for review.

This boundary applies to the digits before the decimal mark, after grouping separators are removed. A value such as 1234567890123456.50 is held as well, because its integer part has 16 digits.

The browser form has no option to lift this boundary. That is intentional: a spreadsheet would round such a number when storing it, so producing it here would only hide the loss.

Reviewing 12345678901234567890

The page example uses Decimal separator . and Grouping separator comma with the source value 12345678901234567890. The expected outcome is that it is kept as text for review: Converted cells does not count it, Values left for review does, and the cell in the output still reads 12345678901234567890.

Being held does not mean the value is wrong. It means the tool will not turn it into a number that a spreadsheet would round.

Decide what the column holds

• Compare the full value with the source, digit by digit.
• Ask what the column means: a reference, account or tracking number is an identifier and should stay text.
• If it is an identifier, exclude it from totals and export it with the text-safe XLSX converter instead.
• If it is a real quantity, such as a very large count, keep the exact digits as text and use software that supports arbitrary precision for any arithmetic.
• Do not try to rebuild missing digits from a rounded spreadsheet value.
• Count the digits rather than reading the value; 15 and 16 digits look alike at a glance.

Related behaviour elsewhere

The decimal totals checker works differently: it accepts long canonical decimals and adds them exactly as whole numbers scaled to the same number of places, so a long value does not lose digits there. That makes it a useful cross-check for a large total, but it is not a reason to store a long identifier as a number.

In the Tablora app, the XLSX export does not apply this boundary. A column that is mostly numeric is written as real numbers, so long digit strings in it would be rounded. Use the app's CSV export for such a column, or move it into the text-safe converter here.

Original synthetic example

Source value

12345678901234567890

Decimal separator

.

Grouping separator

,

Expected outcome

Kept as text for review.

Tools

• numeric-normalizer