How to decide what to automate
first

[Go back]

Automation is easy to get carried away with. You could automate virtually any process, but that doesn’t mean you should. Too much of it gets expensive, time-consuming and a bit soulless.

I only have one coffee a day, so I don’t own a super expensive machine that weighs my beans, grinds them, grabs a cup and serves it to me. That would be stupid. Teak, on the other hand, drinks five coffees a day and hates using a machine. A coffee robot might be a smarter investment for him (but please don’t tell him).

Coffee exaggeration aside, automating a process in your business should start with asking whether you should. Two questions:

[i] Do you have a process that works WELL? Super blunt. Don’t automate an incomplete or half-assed process. You’ll end up spending more time and money than it’s worth.

[ii] How important is the human aspect? Client calls and social events build rapport and keep relationships going. Don’t automate something that should have personality, but you can automate around it.

Now that we’ve got that out of the way, let’s look at the matrix.

Notice the axes: impact vs. repeats.

  • Impact is the cost of the process, in resources, opportunity and failure.
    • Resources: Time and money. That means hard costs like software or contractors, plus how long the process takes someone and what those hours cost. (Are they billable?)
    • Opportunity: What could you or your team be doing, or paying for, instead?
    • Failure: Mistakes happen, and sometimes the consequences are extreme. Who or what pays for it when the process goes wrong?
  • Repeats is how often it happens. Daily, monthly, yearly?

A disclaimer: this is a very first step, and there’s obviously nuance missing. The point is to make a start, so you at least have your most automatable tasks lined up to research further.

Here are a few examples that are plotted on the matrix above.

[i] Every new job needs a proposal, and each proposal takes two days to complete.

Two days is about 14 hours of work. Say this person sends one per fortnight. That’s 364 hours a year. For one engineer at $50/h, that’s $18,200 a year.

[ii] Recording client conversations in the CRM. You sometimes have multiple conversations a day.

Say one engineer makes six calls a week. We don’t automate the call, but we can automate filling in the CRM. That’s 3 hours a week at $50/h, or $7,800 a year, and that’s only one person. It’s about more than the hourly rate, though. It makes sure no one forgets to record a conversation, so no client interaction gets lost in someone’s memory. Very high impact.

[iii] Organising a work Christmas party.

It happens once a year and might take a week if you’re going all out. Building an automation around it would take three times longer. There are more important things to worry about.

Automation should give you time back for the things that matter, like better work, better client relationships, and maybe a second coffee. It shouldn’t be a shiny project that costs more than the problem it solves. Always think about the return on investment for you and your business specifically. Have a go at plotting your own processes on the matrix. If you’d like a hand using it or working out your ROI, hit us up on the for a free, no-obligation chat.