Building an exception playbook ops will actually open

When an API settlement fails at night, nobody wants a twelve-section policy. They want the next action, the owner, and the evidence to capture before the trail ages out.

Keep the spine to five questions

1. What signal opened this exception?
2. Who owns the first response in the next thirty minutes?
3. What must be preserved (raw payload, timestamps, counterpart reference)?
4. When does it escalate to treasury or compliance?
5. What closes it — and who signs that closure?

Write from real tickets

Pull the last month of failed posts, duplicate warnings, and delayed acquirer responses. If a step never appeared in those tickets, it probably does not belong on page one of the playbook.

Taiwan-hour coverage notes

Teams serving regional partners often need explicit language about which roles cover Taipei evenings versus weekend settlement windows. Ambiguity here creates the longest open exceptions we see in reviews.