Workflow logs

Workflow logs is the history of your automations: one row per execution, with the step-by-step trace of what happened.

Open it from the sidebar, or from a rule you are looking at.

The window

The list covers the last seven days, newest first, and so do the counts on the filter pills. “Load older” pages back through that window.

The four numbers at the top are about today, in your own timezone: how many runs, what fraction succeeded, how many failed, and the average duration of the ones that finished.

What a status means

The three filters cover seven states.

Filter	Status	What it means
Success	Success	Every step ran and none failed.
Failed	Failed	At least one step failed. The run may still have finished — see “Continue on error” below.
Running	Starts in …	Queued, waiting out the 60-second undo window on a dashboard write.
Running	Running	Walking its steps right now.
Running	Waiting · resumes …	Suspended at a Wait step.
All	Skipped	Nothing to act on: the record was deleted, or the write that triggered it was undone.
All	Cancelled	The rule was switched off while this run was waiting.

The list refreshes itself every five seconds while anything on it is unfinished, and stops when nothing is. Refresh re-reads everything on demand.

Reading a trace

Each step shows its name, how long it took, and one line of what it did:

Trigger          0ms    record = 4821 · trial, 2026-09-08
Send welcome…    142ms  250 2.0.0 Ok · delivered to jordan@acme.io
Wait 2 days      —      resumes 2026-09-10T09:14:22.000Z
Send the offer…  138ms  250 2.0.0 Ok · delivered to jordan@acme.io
Did they claim…  0ms    took "Otherwise"
Send reminder    121ms  250 2.0.0 Ok · delivered to jordan@acme.io

• A green check is a step that ran.
• A red cross is one that failed; the reason is in the line beneath it.
• A grey dash is a step that was skipped — the branch not taken, or everything after a filter that did not match.
• A timer is a wait, with the instant it resumes.

The duration on the run itself is the sum of the steps’ own time. A run that slept for two days does not report two days: the question the number answers is “how long does this automation take to do its work”.

Failures and “Continue on error”

A run with a failed step is Failed, always. What “Continue on error” changes is whether the steps after it run: with it off they show as skipped, with it on they run anyway — and the run is still red, with the trace showing which step went wrong.

What is in a trace, and what is not

The trace is stored, and admins read it, so the values in it are masked the same way they are everywhere else: a masked column shows as hidden. The run itself reads the real values — it has to, to address the email — but what is written down is redacted.

How long a run is kept

Finished runs are deleted after retention.automationRunsDays (90 days by default). Failed runs are kept twice as long, because the run somebody comes back to read is the one that went wrong. Runs that have not happened yet — pending and waiting — are never swept, however old the row is.

Deleting a rule deletes its runs with it. The confirm dialog says so.

See also: Automations.