The hardest part of automation isn't technical. It's knowing which of the hundred small annoyances in your week is actually worth fixing first. Here's the exercise we'd run if we were sitting at your desk — you can do it yourself this afternoon.
Step one: list the repeats (15 minutes)
Get the people who do the admin in a room — that might just be you — and list every task that's done the same way, more than once a week. Don't judge yet, just collect. Typical entries:
Typing enquiry details into the tracker. Sending the "thanks, we'll be in touch" email. Copying invoices into the spreadsheet. Chasing unpaid invoices. Sending appointment reminders. Compiling the Monday numbers.
If a task needs a decision every time ("is this quote right for this customer?"), leave it off. You're hunting for the same-every-time work.
Step two: put minutes on it (10 minutes)
Against each task, write two numbers: how many minutes it takes, and how many times a week it happens. Multiply. Don't agonise over accuracy — nearest five minutes is fine.
Now the sum that makes it real: weekly minutes × 52 ÷ 60 = hours per year. A 15-minute task done ten times a week is 130 hours a year. At £15 an hour, roughly £1,950 — for one line on your list. (Illustration, as ever: your rates, your volumes.)
Step three: pick one (5 minutes)
Sort by the yearly number and look at the top three. Pick the one that is most boring, most rule-like, and most annoying to the person doing it. Boring and rule-like means it's easy to automate; annoying means the win gets noticed and the next automation gets easier to argue for.
That's your first project. Not a "digital transformation" — one task, moved from a person to a rule.
What NOT to do
Don't start by choosing software — the tool falls out of the task, not the other way round. Don't start with the hardest, most tangled process to "prove the concept"; prove it on an easy one. And be suspicious of anyone whose answer to this exercise is "you need AI" before they've asked what's on your list.
What it looks like when it works
A month later, the task is just… gone. Nobody talks about it. The finance person checks a list instead of typing it. That silence is what good automation sounds like — and it's why the second project is always easier to start than the first.