Merytt Talk through your business
Repeat work

Automate the chasing. Keep the deciding.

Most small firms start automating in the wrong place. Start with the work that happens often, follows the same steps, and leaves someone waiting. Here is how to find yours.

By Aidan SmaleWritten Five minutes to read

Frequentwork that follows the same steps, and has someone waiting on it, is where to start.
6 daysis how long you could spend making a daily task five minutes quicker and still come out ahead over five years, by the xkcd table.
Judgementstays with people. Google’s engineers say work that truly needs it is usually not the thing to automate.
Onejob at a time. Count it before, change it, count it after. Then pick the next.

What should a small business automate first?

Start with work that happens often, follows the same steps each time, and leaves someone waiting. In most small firms that means following up enquiries, setting up new jobs, chasing invoices and sending status updates. Keep decisions with people. Automate one job, check it worked, then move to the next.

That is the short version. The rest of this piece is how to tell which job is yours, and a simple test you can run on your own week.

What are the best things to automate in a small business?

The best things are the jobs nobody would miss doing by hand. They come round every day or week, they are done the same way each time, and when they slip, a client or a colleague is left waiting.

This is our view, from how work moves through small firms. Four jobs come up again and again:

  • Enquiry follow-up. Someone asks for a quote. A reply goes out the same day, and a reminder goes out if they go quiet.
  • Job set-up. A job is won. The folder, the project, the client record and the first invoice get made without anyone typing the same details three times.
  • Invoice chasing. A bill is overdue. A polite reminder goes out on the same day every time, not when somebody remembers.
  • Status updates. A client wants to know where things are. They get a short note when a stage is done, before they have to ask.

None of these is clever. That is the point. They are the jobs that pile up at nine at night, after the real work.

What should you not automate?

Do not automate the calls only you can make. Which client to push on a late bill. Whether a piece of work is good enough to send. What to say to someone who is unhappy. Let a machine prepare those, and let a person decide.

Google’s own engineers draw the same line. In their book on running large systems, they describe the work worth getting rid of as “manual, repetitive, automatable, tactical, devoid of enduring value” and growing as the service grows. Then they add:

“If human judgment is essential for the task, there’s a good chance it’s not toil.”

There is a legal edge too. Under Article 22 of the GDPR, people have the right not to be subject to a decision based solely on automated processing when it has legal or similarly significant effects on them. If a decision really matters to a person, keep a person in it.

Automate the chasing. Keep the deciding.

Is automation worth it for a small business?

Often, yes, but only for the right job. A task you do daily is worth far more effort to fix than one you do once a year. The win is not only time. It is the work getting done on the same day, every time, without anyone chasing it.

A well-known chart by xkcd puts numbers on the trade. It shows how long you can spend making a routine task quicker before you lose more time than you save, across five years. Shave five minutes off something you do daily, and you can spend up to six days fixing it. Shave five minutes off something you do once a year, and you get twenty-five minutes.

We will not tell you how many hours you will save. Nobody has measured that for your business, and we would rather you count it yourself. The chart only makes one thing clear: frequency matters more than anything else.

How do you decide what to automate first?

Score each repeat job on three things: how often it happens, how much it is the same each time, and who is waiting when it slips. Multiply the three. The highest score is your first job. It takes about twenty minutes and a sheet of paper.

This is a free test you can run this week, without us. Write down every job you or your team did more than once last week. Then score each one from 1 to 3 on three questions:

  • How often? 1 for monthly or less. 2 for weekly. 3 for daily or more.
  • How same? 1 if it is different every time. 2 if it is mostly the same. 3 if the steps never change.
  • Who waits? 1 if nobody. 2 if a colleague. 3 if a client, or money.

Multiply the three scores. The top score is 27. Anything at 18 or above is worth a serious look. A job that scores 1 on “how same” usually has a decision inside it, so leave that part with a person.

Two warnings. First, score the job, not the software. Second, be honest about “who waits”. A late invoice reminder is money waiting, even if nobody complains.

How do you know if an automation worked?

Count before you change anything. How many times did the job happen last month, how long did a reply take, and how many slipped? Change one job. Count the same things a month later. If the numbers did not move, fix it or switch it off.

Do one at a time. If you change three things at once, you will not know which one helped. And you will not know which one broke when something goes wrong.

Keep a person watching for the first few weeks. Read what goes out. Check the first ten reminders before you stop looking. The aim is to trust it because you watched it work, not because it was switched on.

If you would like a second opinion on what your first job should be, we are happy to talk it through.

Where this comes from.

Every figure and quote is somebody else’s: published and linked. The four starting jobs and the scoring test are our view, not a study. We do not write numbers we cannot point at.

  1. Eliminating ToilVivek Rau, in Site Reliability Engineering, Google. The six traits of work worth automating, and the note that work needing human judgement is usually not that work.
  2. Is It Worth the Time?xkcd, comic 1205. The five-year table of how long you can spend making a routine task quicker before it stops paying back.
  3. Regulation (EU) 2016/679, the General Data Protection RegulationEUR-Lex. Article 22(1), the right not to be subject to a decision based solely on automated processing.

Last checked 28 October 2026. Nothing here is legal advice. If an automated process makes decisions about people, take advice on your obligations under data protection law.

Questions

What should a small business automate first?

Work that happens often, follows the same steps each time, and leaves someone waiting. In most small firms that means enquiry follow-up, job set-up, invoice chasing and status updates. Automate one, check it worked, then move to the next.

Is automation worth it for a small business?

Often, yes, but only for the right job. A task you do daily is worth far more effort to fix than one you do once a year. The win is the work getting done on the same day, every time, without anyone chasing it.

What should a small business not automate?

The calls only the owner or a senior person can make, such as which client to push on a late bill or whether work is good enough to send. Let a machine prepare those, and let a person decide.

How do I decide which task to automate first?

Score each repeat job from 1 to 3 on how often it happens, how much it is the same each time, and who waits when it slips. Multiply the three. The highest score is your first job.

How do I know if an automation is working?

Count the job before you change it: how often it happened, how long a reply took, and how many slipped. Change one thing, then count again a month later. If nothing moved, fix it or switch it off.

Find your first job with a founder.

Thirty minutes. Tell us what keeps coming round every week, and we will tell you straight whether it is worth automating, and where to start.

Or take the six-question diagnostic