Before You Automate a Task, Ask This First
A simple filter for deciding what's actually worth automating — before spending time or budget on it.
Article illustration — 1600×700px
Automation is easy to start and surprisingly hard to stop once it’s wrong. A bad automation doesn’t just fail to help — it actively creates new work: monitoring it, fixing its mistakes, and explaining to a customer why they got a strange reply.
The three-question filter
- How often does this actually happen? Guess, then check — most people overestimate the frequency of tasks that feel annoying and underestimate the frequency of tasks that feel routine.
- What happens if the automation gets it wrong once in fifty times? If the answer is “nothing serious,” proceed. If the answer involves a customer, money, or trust, keep a human in the loop.
- Who fixes it when it breaks? Automation that nobody maintains degrades silently — decide upfront who owns it before it exists.
The silent failure mode
The most common automation failure isn’t a dramatic error — it’s a slow, unnoticed drift where the automation keeps running technically correctly while the underlying situation has changed, and nobody’s watching closely enough to catch it.
A worked example
A small business considers automating replies to “Are you open on public holidays?” It happens a few times a month (frequency: low-moderate), a wrong answer costs a lost customer or an awkward correction (stakes: moderate), and the answer changes a few times a year around holiday schedules (maintenance: needs a clear owner).
The right call here usually isn’t full automation — it’s a saved, human-reviewed template that gets sent in seconds, with the actual holiday calendar checked once a quarter by someone specific.
The takeaway
The right question isn’t “can this be automated?” Almost anything can be. The right question is whether the frequency, the stakes, and a clear owner for maintenance all line up — and most tasks, honestly assessed, don’t.
Have a topic you’d like covered?