The term comes from finance, where it described trades settling without manual re-entry, and it remains the efficiency ideal for high-volume, rule-bound work. When the rules are truly deterministic, a card payment matching an invoice by reference number to the cent, human review adds cost and delay without adding judgment, and STP is simply correct.
The trouble starts where businesses stretch STP over work that carries judgment: communications, pricing, anything a customer reads. The technology will happily process those straight through too; it just cannot notice what it does not know, and the error rate that is tolerable for internal record-matching becomes intolerable when each error is an email in a customer's inbox.
The mature pattern in small business automation is a split: straight-through for the deterministic middle (matching, syncing, logging) and an approval gate at the boundaries where money or messages leave. Exception handling connects the two; whatever fails the deterministic rules gets kicked to a human queue rather than guessed at. Framed that way, STP and human-in-the-loop are not rivals but layers of the same system, each covering what the other should not.
See the term in practice
Connect your tools read only and watch the approval queue prepare real actions from your own context. Nothing sends until you approve it.