People assume that, as the automation guy, my answer to “should we automate this?” is always an enthusiastic yes.
That would certainly be good for business — building Zapier automations is how I bankroll my insatiable lust for artisanal cheese. But the honest answer is usually “maybe, but not yet”. And I’ve got a four-step framework to explain why.
You’ve probably seen some version of this framework before. It gets applied to everything from achieving inbox zero to deciding which of your children to keep, but this is my automation-specific take.
STEP ONE: Eliminate
Before you automate a process, it’s worth checking whether it needs to exist at all.
Over the years I’ve lost count of the processes I’ve been asked to automate that ultimately turned out to be an inherited habit nobody had ever questioned.
To figure out whether something is worth keeping, ask yourself and/or your team these questions:
- When did anyone last look at the output of this process?
- Who actually asked for this, and do they still want it?
- If this didn’t already exist, would I build it today?
- What breaks if I simply stopped doing this?
If the process can’t justify its existence, then congratulations: you’ve just saved yourself the trouble of automating it.
STEP TWO: Consolidate
(Everyone else calls this step “Simplify”. I am not man enough to have one word in my list not rhyme with the other three. Don’t pretend you wouldn’t have done the same.)
Once a process has earned its place, clean it up before you automate it. Merge or remove unnecessary steps. Standardise how data is collected and structured. Kill the “well, except for…” rule that only exists for that one high-maintenance client nobody wants to upset.
Most people automate a process exactly as it is, without much thought as to whether it could run better or more efficiently. This approach simply bakes in the existing inefficiencies and makes them run faster.
If you sort these issues out ahead of time, your automation will be easier to build, less likely to break, and quicker to fix if it does.
STEP THREE: Automate
As someone who builds Zapier automations for a living, it pains me to say it, but plenty of processes never make it this far.
Eliminate kills some outright. Consolidate shrinks others down so much they’re not worth bothering with. If that happens, brilliant — you’ve solved the problem without building anything.
What’s left is typically the repetitive, rules-based stuff that runs more or less the same way every time. That’s the sweet spot for automation: dull for a person, perfect for a robot.
And thanks to the effort you put into the first two steps, if you do decide to automate you’re now starting from the strongest possible position.
STEP FOUR: Delegate
Whatever’s left after the first three steps is the work that actually needs a human.
This is the work that genuinely can’t or shouldn’t be automated: the creative thinking, the judgement calls, the decisions that require taste rather than rules.
Once you’ve reached this point, it’s time to assign responsibility for the process and the tasks within it. It’s worth noting that delegating doesn’t mean palming it off — the right person for the job could well be you.
Get it written down, too. Map out the steps and, wherever a decision can be reduced to a rule — if this, then that — capture it. The goal is a fully documented process that someone could follow without you in the room.
Final thoughts
So that’s the framework — Eliminate, Consolidate, Automate & Delegate.
And while I’m officially the “automation” guy, in practice the value of the work I do is spread pretty evenly across those four steps.
The cheese budget, alas, depends almost entirely on someone reaching step three and hiring me to do the grunt work — which, coincidentally, is something you can talk to me about by clicking here.