Docs
CSV formats for the free scan
The free scan runs on two files you export yourself - no account connection. This page says exactly what those files need, so you find out before the upload rather than after it.
Shopify orders export
Shopify admin → Orders → Export→ “Export orders” as plain CSV. The standard export already contains everything needed.
Required: a Name column (the #1001 order reference), plus at least one of Total or Created at.
Columns read (everything else is dropped at parse):
NameIdCreated atCurrencyTotalSubtotalTaxesTax amountFinancial statusRefunded amount
Line-item-per-row exports are fine - rows are grouped by order name and the first row carrying a total is used.
QuickBooks Online export
QuickBooks Online → Reports → Transaction List by Date (or Sales by Customer) → Export → CSV.
Required: an amount column (Amount, Total, or Total amount) and a date column (Date, Txn date, or Transaction date).
Columns read:
DateTxn dateTransaction typeNumDoc numberAmountTotalTax amountSales taxMemo/Description
Transaction types counted as sales: Sales Receipt, Invoice, Sale. Counted as refunds: Refund Receipt, Refund, Credit Memo, Credit Note. Any other type is skipped and reported in the scan’s warnings panel, so you can see what was ignored.
What a missing column costs you
The scan never guesses. If a file can’t substantiate a finding, that finding is automatically demoted to needs reviewand the headline figure is recomputed without it - so the number you see is never inflated by data we didn’t actually have.
- No tax column- tax mismatches can’t be confirmed. The Shopify orders export carries order-level tax only, never the line-level split, so tax findings are always needs-review on CSV.
- No refund column- unrecorded refunds can’t be confirmed, because the orders export has no refund date to compare against.
- Sparse QuickBooks reference numbers - orders are linked by the
#1042-style reference in Num or Memo. Without it, only an exact one-to-one amount match can link a row, and anything ambiguous is left unlinked and listed in the warnings.
Limits and formatting
- Up to 5 MB and 20,000 data rows per file. Oversized files are rejected before parsing, not part way through.
- Byte-order marks, Windows line endings, and quoted fields containing commas are all handled.
- Currency symbols, thousands separators, and
(1.23)-style negatives are parsed. Dates may be ISO (2026-03-14) orMM/DD/YYYY.
What LedgerClear refuses to read
Columns are whitelisted, not blacklisted: anything not on the lists above is dropped at parse and never reaches the engine. On top of that, any header matching customer, name, email, phone, address, city, ZIP, province, country, company, vendor, or employee is rejected outright even if it somehow appeared on a keep-list. You can leave those columns in your export - they will not be read.
Uploaded files are parsed in memory and discarded with the request. Only the derived findings persist, and only if you choose to create a shareable report.