Automation is sold as a way to buy back time, and it can be — but most teams reach for it in the wrong order. They automate the flashy thing, or the thing a tool happens to make easy, and end up with a faster version of a process that was never worth speeding up. The skill isn't wiring tools together; it's choosing what to wire together, and in what order. Here's the way I think about it.
Automate decisions you've already made
The best candidates for automation are the tasks where the decision is already settled and only the execution repeats. "When a form is submitted, add the lead to the CRM and send the welcome email" is a decision you make once and then repeat a thousand times. Anything that still requires fresh judgement each time — pricing a non-standard deal, handling an upset customer, deciding strategy — is a poor fit. Automate the repetition, not the thinking.
Rank tasks by time × frequency × friction
Not all repetitive work is worth automating. A task you do once a quarter rarely earns back the effort of building and maintaining the automation. Score your candidates on three things and start at the top:
Time
How long does one round take? Five seconds saved isn't worth a fragile integration.
Frequency
How often does it happen? Daily beats monthly by an order of magnitude.
Friction
How error-prone or draining is it? Tasks people dread or fumble are worth more than the clock alone suggests.
High on all three? Automate it first. Low on all three? Leave it — your attention is better spent elsewhere.
Fix the process before you automate it
Automating a broken process just gives you a faster, more confident way to be wrong. If the steps are unclear, the data is messy, or nobody agrees who owns what, no tool will save you. Map the workflow as it really runs today, cut the steps that exist out of habit, and only then decide what to hand to a machine. Simplify first, automate second.
Start at the seams
The highest-leverage automations usually live at the hand-offs — where work passes between two tools or two people. That's where things get re-typed, dropped, or delayed. Connecting your form to your CRM, your CRM to your email, your invoicing to your accounting: each removes a manual copy-paste and a chance to forget. Look for the seams in your process and stitch those first.
What to leave alone
Knowing what not to automate is half the discipline. Leave these to people:
A warm note to a key partner, a tricky negotiation, the welcome that sets the tone for a new client — these are where the human touch is the value. Automating them doesn't save time so much as quietly erode the thing that made the relationship work.
Keep a human in the loop where it counts
Full automation isn't the only goal — and often shouldn't be. For anything customer-facing or high-stakes, the smart move is to automate the preparation and leave the send to a person: draft the reply, surface the data, queue the action, then let someone glance and approve. You get most of the speed without handing your reputation to a rule you wrote at 2pm on a Tuesday.
Automate the decisions you've already made, fix the process before you wire it up, and leave the judgement and the relationships to people. Speed is only an asset when it's pointed at the right work.
Read next
Want to automate the right things?
I help teams map their workflows and automate the work that actually drains their time — without breaking what makes them good.
Book a Call