Compare
Manual (Excel) or automated validation
In the chain tests of 2025 and 2026 a lot was checked with Excel: open message, add up columns, compare dates. That works for two month-ends, not for twelve months a year across dozens of counterparties.
What manual checking misses
A spreadsheet does not see a missing field, a wrong entityType, a code-list value from another release or a wrongly used switch. It does see totals, and that is exactly what a human is good at. The errors that pass through SIVI's GitHub issues (field lengths, renamed code lists, mandatory fields) are almost all structural errors.
What automation adds
Schema validation in milliseconds, consistency checks on every message, a report with JSON pointers to the error, and an audit trail per month-end. The human remains needed for the exceptions and the coordination with the counterparty.
Our advice
Automate the structural and consistency checks, keep the human on the exceptions. The free validator is the first step; month-end monitoring the second.