Clean the numbers. Protect the identifiers.

A conservative CSV workflow: detect consistent numeric columns, review a representative preview and per-column counts, then download a new file with the original headers and order intact.

Drop a UTF-8 CSV here

Or choose a file up to 5 MB for this browser demo.

Choose CSV

Synthetic sample loaded. No data left this browser.100%

Files are processed locally in this tab and are not uploaded.

Column report

synthetic-orders.csv

3 rows · 6 columns

Ruleset: US-style leading-symbol subset · accounting negatives enabled

order_idProtectedIdentifier-like header; decoded cell values stay unchanged.customerNeeds reviewMixed or ambiguous values; no partial conversion is proposed.amount_usdSuggested3 values can be normalized safely.tax_usdSuggested3 values can be normalized safely.postal_codeProtectedIdentifier-like header; decoded cell values stay unchanged.noteNeeds reviewMixed or ambiguous values; no partial conversion is proposed.

values selected

formula-like flagged

Spreadsheet-safe text hardeningOptional: prefix formula-like text with an apostrophe. Otherwise it exports unchanged and remains flagged.

1 formula-like text cell is flagged and will export unchanged unless hardening is enabled.

ID, postal, account, order, SKU, code, leading-zero, and long-integer columns are protected. Mixed columns stay unchanged for review.

Cleaned preview

Review before export

Row	order_id	customer	amount_usd	tax_usd	postal_code	note
1	0001842	Ada Lovelace	1234.50	123.45	02138	Priority
2	0001843	Grace Hopper	875.00	87.50	M5V2T6	Renewal
3	0001844	Linus Torvalds	2010.00	201.00	94107	=2+3

Showing the first 3 of 3 data rows. Per-column correction counts appear in the report above.

Detect

Only consistent columns qualify.

Preview

Review up to six rows plus per-column correction counts.

Export

A new CSV and audit result.