Map one week of real work.
Pick one area, such as quotes to invoices, new customer onboarding or weekly reporting, and write down every step for a week. Note who does it, how long it takes, and where information is typed, copied or looked up.
You’re looking for the steps people do on autopilot. Those are usually the ones a system can carry.
Score each repeated step.
| Question | Why it matters |
|---|---|
| How often does it happen? | Daily work pays back faster than a yearly task. |
| How long does it take? | Small steps add up when they repeat. |
| Does it follow a clear rule? | Rules can be automated; judgement calls usually shouldn’t be. |
| What does a mistake cost? | Missed follow-ups, wrong totals and late invoices are worth preventing. |
Good first candidates.
- Re-typing between systems. The same details entered in a form, a spreadsheet and an accounts package.
- Reminders and follow-ups. Chasing unpaid invoices, expiring documents or overdue tasks.
- Routine reports. The weekly summary someone builds by hand.
- Forms into records. An enquiry or upload that should create a record, not an email to sort.
Keep people where judgement matters.
Automation should take the repetition, not the relationship. Approvals, pricing exceptions, difficult conversations and anything a customer would expect a person for should stay with a person, with the system bringing them the right information at the right time.
Plan for failure too: what should happen when an automation can’t complete, and who finds out.
Start small, then measure.
Automate one complete workflow, not a scattering of tasks. Record how long it took before and after, and how often errors happened. That evidence tells you whether to go further.
For example, Hero turns a captured request into a reminder and a follow-up on the right day, so chasing stops relying on memory. Invoices shows what’s overdue and drafts the reminder email. A person still decides to send it.