Donworth Studio
Start a project →

What should your business actually automate?

A practical way to identify repetitive work worth removing before adding more software or staff.

Sam Donworth Founder, Donworth Studio 6 min read

The businesses that get the most from automation are not the ones that automate the most. They are the ones that pick the right three or four tasks. Automating the wrong thing costs money, breaks quietly and makes people distrust the next attempt.

Here is the test we use to find the right ones.

The three-part test

A task is a strong candidate for automation when all three are true.

1. It happens often

Daily or many times a day is ideal. Weekly is fine. Something done twice a year is almost never worth automating, however painful it is on the day. Frequency is what turns a small saving into a large one.

2. The rules are clear

Could you write down, on one page, what to do in every normal case? If a cancelled appointment always means the same three actions, that is a rule. If it depends on who the customer is and how the receptionist is feeling, it is judgement, and judgement is expensive to automate and easy to get wrong.

3. It moves information between places

Most repetitive work in a business is not thinking. It is copying: from an email into a system, from one system into another, from a form onto a spreadsheet, from a spreadsheet into a report. Anywhere a person is a bridge between two pieces of software is a candidate.

If a task is frequent, rule-based and mostly copying, it is costing you more than you think and it can almost certainly be automated.

Where it usually pays back first

Across the businesses we work with, the same handful of processes come up again and again.

  • Appointment changes. A cancellation triggers a waitlist message, a calendar update and a note in the customer record. Done by hand, it is ten minutes and often forgotten.
  • Onboarding a new customer or staff member. The same six accounts created, the same welcome email, the same documents sent.
  • Reordering stock. When a count drops below a level, a purchase order goes to the usual supplier.
  • Matching payments to invoices. Bank feed in, invoices out, exceptions flagged for a human.
  • Reporting. The Monday morning figures that someone assembles from four systems.
  • Chasing. Unpaid invoices, unsigned forms, unreturned documents. Polite, timed, and stopped the moment the thing arrives.

None of these are glamorous. Each one typically saves between two and ten hours a week in a small business, and they rarely take more than a few weeks to put in place.

What to leave alone

Some things look automatable and are not, or not yet.

  • Anything involving a complaint. Routing it, yes. Responding to it, no.
  • Processes nobody agrees on. If three staff do it three ways, automating one way will start an argument. Agree the process first.
  • Rare exceptions. Automate the 90 percent that is normal and hand the rest to a person with a clear note. Trying to handle every edge case is where projects stall.
  • Work that is actually relationship. The call to a good customer who is late paying is often worth more than the invoice.

How to find yours

Do not start with a software catalogue. Start with a week.

  • Ask each person to note, for one week, every task they do more than once a day that involves moving information from one place to another. A line each. No detail.
  • Put the lists together and count. The top five will be obvious.
  • For each of the five, estimate minutes per occurrence and occurrences per week. Multiply. Rank by hours.
  • Take the top one that passes the three-part test. Automate that. Nothing else yet.

One working automation that people trust is worth more than five half-finished ones. Once the first has been running for a month, the next one is easy to pick, and staff will start bringing you candidates.

Before adding software or staff

When a business feels stretched, the instinct is to buy another tool or hire another person. Both are sometimes right. But run the week-long exercise first. It is common to find that ten to fifteen hours a week of a role is copying, and that removing it changes the hiring decision entirely.

If you want to see what these automations look like when they are running, our automation page has six of them you can step through with sample data.

Thinking about building something?

Tell us what you need. A rough idea is enough.

Start a project