One consistent invoice shape
Invoice IDs, customers, issued dates, due dates, amounts, currencies, and statuses land in predictable fields.
8 fields
Turn messy AR into records your team can use.
Import a spreadsheet export, normalize invoices into a consistent schema, resolve what needs review, then hand off a clean database-ready CSV.
The small details that keep receivables usable.
Dates repaired
Accepts common spreadsheet dates and writes ISO dates for downstream systems.
Amounts parsed
Currency symbols, commas, and parentheses become numeric amounts.
Rows explained
Missing IDs, customers, dates, or amounts are retained and clearly marked for review.
Export what changed
The download includes the cleaned records, a validation state, and review notes. Nothing is silently discarded.
Bring the raw export.
CSV columns can be named differently. AR Cleanroom recognizes common labels such as invoice number, client, balance, issued, and payment status.
Import
Choose an export or load the representative sample.
Review
Inspect valid records and rows that need attention.
Export
Download the database-ready CSV for your accounting workflow.
No file loaded
0records read
0ready to export
0need review
| Invoice | Customer | Issued | Due | Amount | Status | Review |
|---|---|---|---|---|---|---|
| Load a CSV to inspect normalized records. | ||||||
Source CSV parsing preserves row values and explicit review boundaries
Read the explanation
The sample currency string one thousand two hundred fifty dollars becomes numeric twelve fifty, while parentheses around three hundred fifty point five become negative three hundred fifty point five. Common three tenths pixel per dollar bars extend from one common zero to the right for positive and left for negative. The third sample amount is eight hundred. This is text normalization, not proof that an invoice, balance, currency or debt is valid. The sample contains three data rows. Two have all checked fields and one lacks an invoice identifier. Three equal two hundred pixel row blocks therefore split into two ready and one needs review. The source checks missing identifiers, customers, valid parsed dates and invalid numeric amounts; it retains rejected rows and notes. Ready is a programmed completeness label, not accounting approval, verified collectability or a database import. The actual export has ten columns: seven core invoice fields plus validation state, review notes and source row. Ten equal marks count them, despite a static eight fields marketing label. Source currency defaults USD, status defaults Open, and absent amount text becomes zero because the parser converts an empty stripped string to zero. Last-run storage keeps count and time only; reload loads the sample instead of recovering imported custom rows. No invoice or account service is connected.