How to clean messy bank transaction descriptions

Raw descriptors mix several things in one line: a bank label (POS DEBIT, CHECKCARD), a processor prefix (SQ *, TST*, PAYPAL *, SP *), the merchant name, and codes, dates, phone numbers and locations. Cleaning means keeping the merchant and dropping the rest.

Steps

  1. Export your transactions to CSV. If you only have a PDF statement, RowSmith converts it to CSV in your browser.
  2. Remove bank labels such as POS PURCHASE, POS DEBIT, CHECKCARD and ACH DEBIT.
  3. Remove processor prefixes and note the processor if you care about it.
  4. Remove store numbers, reference codes, phone numbers, and trailing city and state.
  5. Fix capitalisation, then group duplicates under one name.
  6. Assign a category by merchant, and check anything that matters for taxes yourself.

Doing this by hand or with spreadsheet formulas works for a few dozen rows. For thousands, or inside an app, an API call per batch is less work.

Examples

Example lines with made-up merchants, and what the Merchant Cleanup API returns for them today:

Raw lineCleaned merchantProcessorCategory
POS PURCHASE SQ *MAPLE STREET COFFEE 1234 OAKLAND CAMaple Street CoffeeSquaremeals
TST* JOES DINER #0042 AUSTIN TXJoes DinerToastmeals
CHECKCARD 0512 RIVERSIDE FUEL 57444 SAN JOSE CARiverside Fuel-auto_and_transport
PAYPAL *PINE HILL BOOKSTORE 4029357733 CAPine HillPayPalother

Results are automated suggestions and can be wrong; check anything that matters.

FAQ

Can I do this with spreadsheet formulas?

Yes, for simple cases: SUBSTITUTE and REGEXREPLACE (in Google Sheets) can strip known prefixes and digits. It gets hard once you have many formats.

How accurate is automated cleanup?

It varies. Results are suggestions; small and local merchants are the hardest. The API returns a confidence score and how it reached each result so you can review the low ones.

Do you store the descriptors I send?

No. Descriptors are processed in memory and not saved by us. See the privacy page.

Try it

Try this line in the free demo Get a free API key

Have a PDF bank statement instead of a CSV? RowSmith converts bank statements to CSV or Excel in your browser, then you can clean the descriptions here.

More guides: What does SQ * mean on a bank statement? - What does TST* mean on a bank statement? - What does PAYPAL * mean on a bank statement? - What does SP * mean on a bank statement? - What does POS DEBIT or POS PURCHASE mean on a bank statement? - What do ACH and CHECKCARD mean on a bank statement?