The Cost of Two Copies
Every payment system eventually builds a reconciliation operation.
It may be a small team with a spreadsheet, a nightly job that matches files, or an entire department armed with exception queues and carefully named CSV exports. The tooling varies; the underlying problem does not. A single economic event is captured independently by several systems, and somebody must later decide whether their records describe the same thing.
That is an expensive way to learn the truth.
A merchant has an order. A processor has an authorization. An acquirer has a clearing record. A bank has a movement. A wallet has an on-chain transfer. Each record is useful, each is locally correct, and none is automatically the other. Reconciliation is the work required to turn those separate claims into one operational fact.
The usual answer is more matching: stronger references, better files, more frequent imports, smarter exception handling. Those are worthwhile improvements, but they optimize the aftermath. They do not change the fact that the system started with multiple copies.
Counterpart begins somewhere else: make the relationship between a ledger entry and the thing outside the ledger explicit from the beginning.