Stop-on-reply
The moment a contact replies, their sequence stops — no more follow-up steps are sent to them, even if later steps are still scheduled.
How it works
SubsiMail checks every connected mailbox for replies every minute — whether or not a campaign is running, because people answer days or weeks after a sequence ends. Only mail that answers an email SubsiMail sent counts: it is matched by the message's reply headers, so a conversation you started yourself from your normal inbox, or a fresh email from the same person, is ignored. (A message without reply headers counts only if its subject starts with “Re:”.)
When a reply is found the contact's status becomes replied and no more steps are sent to them; every other contact carries on. The reply is stored and shown on the contact's page and in the replied-events list, where the eye icon opens it without leaving the page. Out-of-office and other automatic replies also stop the sequence, but are never answered by the auto-reply.
Replies are read from the mailbox the campaign's Reply-To points at, so a mailbox that can only send (such as Amazon SES) should use a Reply-To that is a real inbox connected with IMAP. The Accounts page lists, per mailbox, where each campaign's replies go and warns for any campaign whose Reply-To can't be read.
How detection works per provider
| Gmail | Checks the connected inbox for new messages from the contact's address since the sequence started |
| Outlook | Same idea, scoped specifically to the Inbox folder (not Sent Items) |
| SMTP/IMAP | Reads only the mail that arrived in the IMAP inbox since the last check, so it stays fast however big the inbox is. The first check of a mailbox looks back 30 days. |
Why a self-test doesn't trigger it
If you test a campaign by enrolling yourself and your sending account happens to share an inbox with the contact address, reply detection won't misfire on your own outgoing message. Outlook's check is deliberately scoped to the Inbox folder specifically (excluding Sent Items) for exactly this reason, and a short buffer after send time is applied across providers before a "reply" is trusted — both guard against a real self-send test being mistaken for an actual reply.