Before you automate: a five-question workflow audit
Automating a broken process makes it break faster. Five questions to decide which workflows to automate, which to fix first and which to keep manual.
Automation is having a moment. Every tool promises to connect your apps, write your emails and run your operations. Some of that promise is real. But the businesses that get value from automation tend to do one thing first: they look closely at the process before they automate it.
Here’s a short audit you can run on any repetitive workflow.
1. Can you describe it in steps?
Write the process down as a numbered list, from trigger to finish. “When a new enquiry arrives, someone reads it, decides who should handle it, replies, and adds it to the spreadsheet.”
If you can’t write it down, or if three team members would write three different versions, it isn’t ready for automation. Agree on the process first.
2. How often does it happen, and how long does it take?
A task that happens twice a year rarely justifies building an automation. A task that happens twenty times a day, even if it only takes a few minutes, usually does. Estimate frequency and time honestly, ideally by having the team log it for a week.
3. Where does it need judgement?
Mark every step that requires a decision. Some decisions follow clear rules (“if the topic is pricing, send it to sales”), and these can be automated. Others need context, empathy or accountability (“is this complaint serious enough to call the customer?”), and these should stay with a person, perhaps supported by an AI summary.
Most good workflows are a mix: automated steps with human checkpoints at the moments that matter.
4. What happens when it goes wrong?
Every automation will eventually meet an input it doesn’t expect. Before building, decide:
- How will you know something failed?
- Who gets alerted?
- What’s the manual fallback?
If you can’t answer these, the automation could fail silently, which is often worse than not automating at all.
5. Where does the data live?
Automation moves information between tools. Check that the tools you use can actually connect (through native integrations, an automation platform or an API), and think about what data is being moved. Customer details deserve care: only move what’s needed, and know where it ends up.
Scoring the result
After the audit, each workflow tends to fall into one of three groups:
- Automate now: clearly defined, frequent, rule-based, with a known fallback.
- Fix first: valuable, but the process is inconsistent or unclear.
- Keep manual: rare, high-judgement, or high-risk.
Start with one workflow from the first group. Build it, monitor it for a few weeks, and only then move on. A small number of reliable automations beats a large number of fragile ones.
If you want a second pair of eyes on your processes, our AI automation service starts with exactly this kind of discovery. You can also see a workflow run live before you decide, or try the Growth Leak Finder to see where to look first.