1. Pull Ramp transactions for the month. Amount, merchant, date, cardholder, and GL category for every transaction in the period.
2. Pull the NetSuite journal entries. The corresponding booked entries from NetSuite for the same date range.
3. Match on amount, merchant, and date. Parabola's join and fuzzy match line up rows even when merchant names don't agree, so "Amazon," "AMZN MKTP," and "Amazon.com" resolve to the same vendor.
4. Surface the gaps in both directions. Anything in Ramp with no NetSuite entry is not yet booked. Anything in NetSuite with no Ramp match is a potential ghost entry.
5. Output a clean exceptions report. Only the rows that need a human, grouped by gap type, dropped into a Parabola Table or sent to the close team.