Automations are easy to build and easy to forget. You connect a form to a spreadsheet, a spreadsheet to a CRM, a CRM to an email tool, and for months everything seems fine. Then one day a client tells you they never got the welcome email, and you discover the workflow broke weeks ago when someone renamed a column.
This guide gives you a simple quarterly audit you can run in an afternoon. You will learn how to build an inventory of every automation you own, how to spot the ones that fail quietly, how to decide what to keep, fix or delete, and how to document the survivors so the next review takes half the time.
Why automations need a regular review
Most automation platforms are very good at running a workflow and much less good at telling you that the workflow no longer makes sense. A trigger can keep firing perfectly while the data it moves has become useless. Over time, a few predictable problems pile up:
- Silent failures. A step errors out, the platform logs it, and nobody reads the log. Error emails often land in a folder no one checks.
- Schema drift. A field gets renamed, a dropdown gets a new option, or a spreadsheet tab gets moved. The automation still runs but maps data to the wrong place.
- Orphaned workflows. The person who built a workflow changes roles or leaves, and the automation keeps running under an account nobody watches.
- Duplicated logic. Two people solve the same problem in two tools, so a new lead gets two welcome emails or is added to the CRM twice.
- Permission creep. Old connections keep access to inboxes, drives and calendars long after they stopped being needed.
None of these problems is dramatic on its own. Together they make your systems less trustworthy, and once people stop trusting an automation they start doing the work by hand again, which defeats the purpose.
Step 1: Build a complete inventory
You cannot audit what you cannot see. Start by listing every automation in one place, even the small ones. A plain spreadsheet or a simple database in your note-taking app is enough. Create one row per workflow with these columns:
- Name and a one-sentence description of what it does in plain language.
- Platform where it lives (for example Zapier, Make, Power Automate, a native app integration or a custom script).
- Trigger: the event that starts it.
- Apps touched: every tool it reads from or writes to.
- Owner: the person responsible for it today, not the person who built it years ago.
- Business purpose: what would go wrong if it stopped.
- Last reviewed date.
To find everything, check each automation platform your team uses, then look at the integrations or connected apps page inside your main tools: your CRM, email platform, project manager, calendar and file storage. Native integrations are the ones people forget most often because they do not live in a dedicated automation tool. Also ask the team a direct question: “Is anything sending, copying or creating data for you automatically that is not on this list?” You will almost always discover one or two.
Step 2: Check health and run history
With the inventory ready, go through each workflow and look at its recent activity. Most platforms keep a task or run history. For each row, note three things:
- When did it last run successfully? A workflow that has not run in months is either broken or no longer needed.
- How many errors did it have since the last review? Even a handful of errors deserves a look, because each one may represent a lost lead, a missing invoice or an unsent reminder.
- Is the output correct? Open the destination and check a few recent records by hand. Make sure the right fields are filled, dates are in the right format and nothing is landing in a column that was renamed.
That third check is the one people skip, and it is the most valuable. An automation can show a perfect success rate while writing phone numbers into the email field. Pick three to five recent records per workflow and compare them with the source. If you find a mismatch, mark the row in red and move on; you will fix it in the next step.
Test the failure path, not just the happy path
For your most important workflows, deliberately trigger an edge case in a safe way: submit a test form with an empty optional field, a very long name or an accented character. See whether the workflow handles it gracefully or stops. If it stops, find out who would be notified. If the honest answer is “nobody,” that is a finding in itself.
Step 3: Decide to keep, fix, merge or retire
Now sort every workflow into one of four buckets. Being strict here is what makes the audit worth the time.
- Keep: it runs cleanly, the output is correct and someone still needs the result. Update the last reviewed date and move on.
- Fix: it is still needed but has errors, wrong mappings or a missing owner. Write down the specific fix and who will do it, with a date.
- Merge: two or more workflows do overlapping jobs. Choose the one that is easier to maintain, move the missing steps into it and turn the other off.
- Retire: nobody can explain why it exists, or the process it supported has changed. Turn it off first, wait a couple of weeks to see if anyone notices, and then delete it.
The “turn off before deleting” rule protects you from removing something that quietly mattered. Many platforms let you pause a workflow instead of deleting it, which keeps the configuration available if you need to bring it back. Add a note in the inventory with the date you paused it and the date you plan to delete it.
Step 4: Clean up access and connections
Every automation needs permission to read or write somewhere, and those permissions tend to outlive the workflows that used them. As part of the audit, review the connected accounts in each automation platform and in the security or connected apps settings of your main tools.
- Remove connections that are not used by any workflow still marked keep or fix.
- Reconnect workflows that run under a personal account of someone who has left, ideally using a shared service account your team controls.
- Check whether each connection asks for more access than it needs. A workflow that only reads a calendar should not have permission to delete events if a narrower option exists.
- Rotate API keys and webhook secrets that were shared in chat messages or documents, and store them in your team’s password manager instead.
This part of the review is not glamorous, but it reduces risk in a way that is easy to explain to anyone who asks why the audit matters.
Step 5: Document what survives
The goal of documentation is not a perfect manual. It is to make sure anyone on the team can understand a workflow in two minutes. For each automation you keep, add a short note to the inventory or in the description field of the workflow itself:
- What starts it and what it produces, in one or two sentences.
- Any assumptions it depends on, such as “the Status column must keep the value Approved.”
- Where errors are reported and who reads them.
- How to pause it safely if something goes wrong.
Naming also helps. A consistent pattern such as “Source to Destination: purpose” (for example, “Website form to CRM: new lead”) makes the list scannable and makes duplicates obvious at a glance.
Making the next audit faster
The first audit is the slowest because you are building the inventory from scratch. To make the following ones quicker, build a few habits into your normal work:
- Add every new automation to the inventory the day you create it, with an owner.
- Send error notifications to a shared channel instead of a personal inbox, and agree on who checks it.
- When you change a form, a spreadsheet structure or a CRM field, search the inventory for workflows that touch it before you save the change.
- Put the next audit on the calendar now, with a recurring reminder every three months.
A simple example: a small agency with around twenty automations ran its first audit and found three workflows that had been failing for weeks, two that duplicated each other and four that nobody needed anymore. After cleanup they had eleven workflows, each with a named owner. Their next review took less than an hour.
Summary
A quarterly automation audit keeps your systems reliable without much effort. List every workflow with its trigger, apps, owner and purpose. Check the run history and, more importantly, inspect real output records. Sort each workflow into keep, fix, merge or retire, and pause before deleting. Remove unused connections and move critical workflows to shared accounts. Finally, write a two-minute description for each survivor and schedule the next review. Done regularly, this turns a fragile collection of automations into a system your team can actually trust.
Photo: Person writing in notebook while using laptop at a modern workspace, by Shixart1985, license CC BY 2.0, via Wikimedia Commons. Cropped.