How to choose what to automate first

Why the first automation you choose matters most

Picture a café owner on a Sunday night. The roster's half done, there's a supplier order to get in before Monday, and the phone keeps buzzing with shift swap messages.

Every one of those jobs could be automated. So which one goes first?

It matters more than you'd think. Your first automation sets the tone for everything that follows. If it saves real time and runs quietly in the background, your team will trust the next one.

If it's ambitious, half-finished or changes how everyone works overnight, it becomes the project nobody wants to talk about.

That's why we start with one task. Not a platform, not a transformation programme, just one piece of repetitive work that goes away. You see it running on your real jobs and real customers, and only then decide what to do next.

Step 1: List the repetitive work

Spend fifteen minutes writing down every task that happens again and again. Don't filter yet. Just get it all on paper.

Ask the people who actually do the work. Owners often underestimate how much admin lands on someone else's desk, and the person doing it usually knows exactly where the time goes.

Good prompts to get the list going:

• What do you type into more than one system?
• What do you copy and paste from emails, texts or forms?
• What do you chase every week: documents, payments, confirmations, timesheets?
• Which questions do customers ask over and over?
• What report or summary do you put together by hand?

The answers tend to look familiar. Rosters and shift swap messages in cafés. Quotes written up from site notes in trades. Booking reminders in clinics. Stock counts and reorders in shops. Chasing client documents for advisers. Filing proof of delivery for transport companies.

If your list has more than a dozen items, that's normal. You're only picking one.

Step 2: Time each task for one normal week

Before you decide anything, measure. Gut feel is a poor guide here: the job that feels endless might take an hour, and the one you barely notice might quietly eat a whole afternoon.

Pick a normal week, not the week before Christmas, and keep a rough tally of how long each task takes. A note on your phone is enough. Nobody needs a stopwatch.

Our typical estimates for the top three tasks in the industries we work with range from about six and a half to nine hours a week. Those are estimates, not results, and your own tally is what counts.

Step 3: Score each task to find what to automate first

Now you've got a list and some real hours. Score each task from 1 to 5 on four things, add the scores up and sort the list.

The highest score is your first candidate. It won't always be the job you expected.

If two tasks tie, pick the one your team will notice most. A visible win builds momentum, and that makes the second automation much easier to get going.

Step 4: Watch for these warning signs

Some tasks look like good candidates on paper but are better left for later. Check your top scorer against these before you commit:

• The process isn't settled. If everyone does it differently, agree on one way of working first. Automating a messy process just makes the mess faster.
• It depends on a decision. Automate the preparation, such as gathering information or drafting a message, and keep the decision with a person.
• It involves sensitive personal information. It can still be automated, but check how the data is stored, who can see it and where it goes. Our Privacy Act guide covers what to check.
• Nobody owns it. Every automation needs one person who notices if it stops working.

A warning sign doesn't rule a task out forever. It just means there's a bit of groundwork to do first, and your second-highest scorer might be the better place to start.

Step 5: Start with one task, then measure the hours saved

Once you've chosen your first task, the temptation is to bolt on "just one more thing". Resist it. Keep the scope tight:

1. Write down what "done" looks like. For example: "every approved job creates a draft invoice in Xero within a few minutes, with the right customer and line items."
2. Build it around the tools you already have. New software should be the exception, not the starting point.
3. Run it alongside the old way. Check its work on real jobs for long enough to trust it before you switch the manual process off.
4. Measure the time saved. Compare against the week you timed in step 2. That number tells you whether to do the next one.

That last step is why the timing in step 2 matters so much. Without a starting figure, you're left guessing whether it worked.

What a good first automation looks like

Good first automations share a few traits. They take a task that's repetitive, rule-based and frequent. They connect one or two tools you already use. And their output is easy to check.

Here are some common examples:

Notice what they have in common. Each one replaces the same small steps done over and over, and if something looks off, a person can spot it straight away.

Not sure which task to automate first?

Back to that café owner on a Sunday night. With a list, a week of timings and a simple score, the choice stops being a guess. One job goes first, and the rest wait their turn.

If you're still not sure which of your tasks fits, that's exactly what our free session is for. We sit with whoever does the work, time it with you, and leave you with a written plan and an hours estimate that's yours to keep.

Common questions