Problem A08
My automation broke and I did not find out until someone complained.
No link on this page is an affiliate link. We are in the programme below, so we could earn a commission elsewhere on this site — but not from anything on this page. We recommend free tools where they are the better answer, including ones that earn us nothing.
- Make.com: 35% of subscription payments for 12 months; extra-operation purchases do not earn commission (terms verified 2026-09-24)
What you are doing now
You assume it is working. Days later you discover it has not been, and you reconstruct by hand what was missed.
How often: Occasional, but it is the single biggest reason people abandon automation.
What you want instead
A failure notifies a human immediately, retries where it is safe to do so, and leaves a record of exactly what did not run.
The free option first
A manual weekly spot-check costs nothing and is genuinely the right answer for low-stakes workflows.
Who should not do this
Nobody should skip this. If a workflow matters enough to build, it matters enough to know when it stops.
How to stop this happening
Every item below exists because we measured the failure it prevents, not because it is general advice.
- Put the alert BEFORE or OUTSIDE the step that can fail. In our run the notify step sat after the failing step, so it never executed and nobody was told.
- Do not rely on a default alert. On all three platforms we tested, no alert reached anyone by default.
- Mark the record, not just the run. Our partial failure wrote 4 of 10 rows and nothing flagged the order as incomplete.
- Reject empty payloads explicitly. Our runner accepted an empty payload silently and carried on.
- Fail the send when a merge field is missing. Ours delivered a message containing a literal {{FirstName}} to the recipient, with no error raised.
- Decide what a retry is worth before enabling it. Adding retries took our failures from 6 to 4 and consumed 14 extra attempts.
Before you build anything for this, the Automation Decision Sheet works out whether it is worth automating at all. It is free, needs no email address, and often concludes that it is not.