How to spot the tasks worth automating

Automation projects fail for a predictable reason: someone automates the task that was most irritating rather than the task that was most costly.

Irritation and cost are barely correlated. The three-minute job you do forty times a month costs far more than the ninety-minute job you hate doing once a quarter.

Here's a way to tell them apart.

Start with an honest inventory

For one week, log what you and your team actually do in thirty-minute blocks. Not what the job description says — what happened. Use whatever's fastest: a notes app, a shared sheet, voice memos.

You need frequency and duration. Everything else follows.

Score each task on four dimensions

Frequency. How many times per month? This is the biggest multiplier and the most under-weighted.

Duration. Realistic minutes, including the context-switching cost. A two-minute task that interrupts deep work costs more than two minutes.

Rule clarity. Could you write the decision logic on one page? If the rules change based on judgment, tone or relationship, automation will be brittle.

Error cost. What happens when it's done wrong? High-error-cost tasks are excellent candidates — machines don't get tired at 6pm — provided the rules are clear.

Multiply frequency by duration for monthly time cost. Then filter by rule clarity. Anything with high time cost and clear rules goes to the top.

The tasks that consistently score well

• Moving data between two systems
• Sending predictable notifications and reminders
• Creating records — tickets, invoices, checklists, folders
• Routing and assignment based on fixed criteria
• Scheduled reporting
• Standard follow-up sequences

Notice the pattern: these are all connective tissue between decisions, not decisions themselves.

The tasks that look automatable but aren't

• Anything requiring negotiation or judgment about people
• Processes with more exceptions than standard cases
• Work that changes every month — automate a moving target and you'll maintain it forever
• Tasks where the human contact is the value
• Anything done once a quarter, no matter how tedious

Simplify before you automate

Half of what looks like an automation opportunity is actually an elimination opportunity. For each candidate, ask in order:

1. 1Can we stop doing this entirely? Who would notice?
2. 2Can we do it less often — weekly instead of daily?
3. 3Can we remove steps?
4. 4Can we standardise it so there's only one version?
5. 5Now — can we automate it?

Skipping straight to step five is how businesses end up with elaborate automations for processes that shouldn't exist.

Estimate honestly

Rough test: if a task costs four hours a month and the automation takes eight hours to build plus an hour a month to maintain, you're ahead within three months. That's a good investment.

If it costs thirty minutes a month, it will never pay back — no matter how satisfying it would be.

Include maintenance in the estimate. Every automation is a small ongoing commitment: it breaks when a tool updates, when a form changes, when a process shifts. Five well-chosen automations you maintain beat twenty you don't.

Where to begin

Take your list, sort by monthly minutes, filter to clear rules, and build the top one. Run it for two weeks. Then build the next.

One working automation teaches you more about your operations than a month of planning.