What’s Fixed
When an automation step failed, the whole run stopped there. Every step after it — send an email, add a tag, create a task — silently never ran, and the only sign was a run marked Failed. For the Assign step that was the wrong trade: the problem it was fixed for was that the step lied about having assigned the contact, not that the run carried on past it.
A lost assignment no longer stops the rest of the automation
If an Assign step cannot make the assignment stick — the write is discarded, or the round-robin group hands the lead to nobody — the step reports Failed and says why, and the rest of the automation still runs. The emails, tags and tasks after it happen as the operator designed them.
This applies only to an assignment that did not take effect. A misconfigured step — a group, user or pond that no longer exists, or a step with no target at all — still stops the run, because nothing after it can be trusted to be acting on the contact the operator meant.
The run itself reports the failure
A run that continued past a failed step no longer reports Completed. Once the automation has been walked to its end, the run ends Failed, names the step’s reason, and counts toward the automation’s failure count — so it shows up where an operator looks, instead of disappearing into a green run.
While the run is still going, the reason is shown as an in-progress warning rather than a red error box, so a run that is still working is not mistaken for a dead one.
Retry actually retries
Clicking Retry on a failed run now re-attempts the steps that failed, and picks the run back up from there:
- The failed step is genuinely re-run — previously the run could come back Completed with nothing re-attempted and the contact still unassigned.
- If the failure stopped the run, Retry carries on through the rest of the automation once the step succeeds. The emails, tags and tasks the failure skipped finally happen, instead of the run reporting Completed with them never run.
- Steps whose last recorded outcome was a success are not re-run, so a Retry never re-sends an email or re-creates a task. A step is judged on its most recent attempt, so a step inside a repeating loop that worked one time round and failed the next is still re-run.
- Each failed step gets exactly one attempt per click, and the run is judged on that attempt alone — a retry that fixes the cause reports Completed, and one that fails again reports Failed once rather than counting the same failure twice.
A failed run no longer locks the contact out of the automation
An automation set to re-enroll a contact after completion used to be permanently closed to any contact whose run had failed: the enrollment was left open forever, and the automation’s own rule then refused to enroll them again. A failed run now closes its enrollment, so the contact is eligible the next time the automation’s trigger fires — and clicking Retry reopens that same run rather than starting a second one.