The worst automation failures are the quiet ones. A form changes a field name, the integration stops passing leads into the CRM, and nobody notices for three weeks until someone asks why enquiries have dried up. Automation monitoring is the set of checks, alerts and owners that catch these failures in hours instead of weeks. This article explains why automations break, what to monitor, and a simple ownership model a small team can run without a dedicated operations engineer.
Why automations break
Automations sit between other systems, and those systems change without asking you. Common causes include:
- An app changes its API, a field name or a form layout.
- A login expires, a password changes or the person whose account powered the connection leaves the company.
- A plan limit is reached, such as tasks, credits or executions, and runs pause.
- Unexpected data arrives: an empty field, a phone number in a new format, a currency the workflow did not expect.
- Someone edits the workflow to fix one thing and breaks another.
- The process changes in the business, but the automation does not.
Some of these produce an error message. Many do not. A workflow that runs “successfully” while writing blank values into your CRM looks healthy on the platform’s dashboard.
Step 1: build an automation register
You cannot monitor what you have not listed. Create a simple register (a spreadsheet is enough) with one row per automation:
- Name and a one-line description of what it does
- Trigger and the systems it touches
- Owner (one named person) and a backup
- Whose account or API key it runs on
- How often it should run, such as “every new order” or “daily at 7 am”
- What breaks in the business if it stops
- Where its documentation and process map live
Rank each automation as critical, important or minor based on that “what breaks” column. Critical ones get the strongest monitoring.
Step 2: set up three layers of automation monitoring
Layer 1: error alerts
Every platform can notify you when a run fails. Switch this on for every automation and send alerts to a shared channel or inbox that the owner and backup both watch, not to the personal email of whoever built it. Zapier, Make and n8n all log failed runs, and Make and n8n let you add error-handling routes inside a workflow, for example to retry, skip a bad record or send a detailed alert.
Layer 2: silence alerts
Error alerts do not catch a workflow that simply stops being triggered. For critical automations, add a check for missing activity: “If no new lead has reached the CRM in 24 hours on a weekday, alert the owner.” Some monitoring services offer this as a heartbeat check, where the workflow pings a URL each time it runs and you are alerted if the ping stops. You can also build it as a small scheduled automation that counts yesterday’s records.
Layer 3: output checks
The quietest failures produce output that is wrong. For important workflows, check the results:
- A daily count comparison, such as orders in the shop against invoices created in accounting
- A check for blank or default values in fields that should always be filled
- A weekly sample of five records reviewed by a person
Say a Lucknow coaching institute automates moving fee payments from its payment gateway into its student records. Error alerts are on. One month the gateway adds a new payment method, and those payments arrive with a different field structure. The workflow runs, but writes blank student IDs. A daily check comparing gateway payment totals with recorded payments flags the gap the next morning, before parents start receiving wrong reminders.
Step 3: assign ownership properly
Every automation needs one named owner. The owner does not have to be technical; they need to understand what the automation is for and be the person who notices when the business result looks wrong. Their duties:
- Respond to alerts within an agreed time: same day for critical, within a few working days for minor.
- Approve changes to the workflow before they go live.
- Review the automation when the underlying process changes.
- Hand over to the backup when on leave.
Run connections on shared service accounts or a business admin account where the tools allow it, rather than a staff member’s personal login. When someone leaves, check the register for anything running on their accounts before you close them.
Step 4: control changes
Well-meant edits are a common cause of breakages. A few simple habits reduce the risk:
- Duplicate a workflow and test changes on the copy before editing the live version.
- Write a one-line note in the register for each change: what, why, who and when.
- Use the platform’s version history where available, so you can roll back.
- Test with real examples, including at least one awkward case, before switching over.
Step 5: review monthly
Once a month, spend 30 minutes on the register. Check failed runs, plan usage against limits, upcoming login or API key expiries and any automation nobody has looked at in months. Retire automations that no longer serve a purpose. Every unused workflow is one more thing that can break or leak data.
A monitoring checklist
- Every automation is in the register with an owner and a backup.
- Error alerts go to a shared channel, not one person’s inbox.
- Critical automations have a silence alert or heartbeat.
- Important automations have an output check.
- Connections run on business accounts, not personal ones.
- Changes are tested on a copy and logged.
- Plan limits and expiry dates are reviewed monthly.
We add monitoring, registers and handover documents to everything we build in our automation projects, and the client owns all of it at handover. For teams that need ongoing technical oversight without a full-time hire, our part-time CTO support can cover it. If you suspect some of your automations are failing quietly, let us know on a call.
Frequently asked questions
How do I know if a Zapier or Make automation has stopped working?
Turn on the platform’s failure notifications and send them to a shared channel. Then add a check for missing activity, such as an alert if no new records arrive in a set period, because a workflow that stops triggering may not raise any error. For important workflows, compare counts between the source and destination systems daily.
Who should own business automations in a small company?
The person responsible for the business result the automation supports, such as the sales lead for lead routing or the accounts person for invoicing, with a named backup. They do not need to build workflows themselves, but they should receive alerts, approve changes and notice when results look wrong. Technical help can sit with an internal builder or an outside partner.
What is a heartbeat check?
A heartbeat check confirms that something is still running. Your workflow sends a small signal, often a request to a monitoring URL, each time it runs. If the monitoring service does not receive the signal within the expected interval, it alerts you. It catches workflows that silently stop, which ordinary error alerts miss.