A small business can automate the boring, repeated jobs: anything that happens every day or every week, where someone currently checks a screen or a spreadsheet by eye, where the data already lives in a system you're paying for, and where a wrong answer would be obvious quickly. For a lot of small firms that means a morning summary of the day's work, chasing overdue invoices, nudging quotes that have gone quiet, or pulling numbers from three places into one. What you should leave alone is anything that needs judgement, conversations with customers that need a human tone, processes that change every month, and jobs that only happen twice a year. And before building anything, check whether the software you already pay for does it, because it often does.
That last point matters more than it sounds. A fair chunk of what gets called "automation" is a setting someone never switched on.
What makes a good first automation?
There's a simple test, and a job has to pass all four parts of it to be worth automating first.
It happens daily or weekly. Frequency is where the value comes from. A ten-minute job done every morning is around forty hours a year. A ten-minute job done once a quarter is forty minutes. You can automate both, but only one of them pays for itself.
Someone checks it by eye. The best candidates are the jobs where a person opens a system, looks at a list, and decides who needs to hear about it. Who's overdue? Which jobs are on today? Which quotes haven't had a reply? If the job is "look at a list and tell someone", it's an automation waiting to happen, because looking at a list is exactly what software is good at.
The data already lives in a system you pay for. Your accounts package, your booking system, your Outlook calendar, a SharePoint list, the CRM. If the information is already typed in somewhere, an automation can read it. If the information only exists in someone's head or on a whiteboard in the warehouse, the first job isn't automation, it's getting it written down.
A wrong answer is cheap to spot. For your first one, pick something where a mistake is visible and harmless. A morning email that lists the wrong job will get noticed by about half past seven, and nobody's hurt. An automation that quietly changes prices in your accounts is a different animal. You want your first automation to build trust, and trust comes from being able to see it working.
If a job passes all four, it's a good first build. If it fails one, it might still be worth doing later, but it isn't where you start.
Does your software already do it?
Check this first, every time. It's the cheapest automation there is.
Xero, QuickBooks Online and Sage Accounting all have built-in overdue invoice reminders. In Xero you switch them on from the invoices screen and choose how many days overdue each one goes out. QuickBooks Online lets you set up to three automatic reminders before or after the due date. Sage Accounting has an automatic payment reminders option in its settings. If polite chasing on a schedule is all you need, switch it on and keep your money. The overdue invoice reminders guide goes through when the built-in version is enough and when it isn't.
The same is true elsewhere. Salon booking systems such as Fresha send appointment confirmations and reminders by text and email on their own. It's worth a look through the settings of whatever job or booking app you use before assuming it can't do something. Microsoft 365 can already forward emails by rule, and Outlook has scheduled sending built in.
It's only worth building something when the built-in feature can't do what you need: the reminder has to go to the account manager rather than the customer, or the information you need is spread across two systems that don't talk to each other.
What does that look like in different businesses?
The test stays the same; the candidates change with the shape of the business.
For an owner-manager, the first win is usually visibility. You check the bank, the accounts, the inbox and the diary every morning to work out where you stand. A single email at 7am with today's jobs, yesterday's takings and anything overdue replaces that loop. That's the daily business summary email, and honestly it's the one I'd suggest first to almost anyone, because it passes all four parts of the test and everyone can see whether it's right.
For a salon, the client-facing reminders are almost certainly in the booking system already. The gaps tend to be behind the scenes: a weekly summary of no-shows and cancellations sent to the owner, a list of clients who haven't rebooked in eight weeks for whoever does the follow-up calls, or the day's till takings dropped into the spreadsheet the accountant wants.
For trades, the obvious one is quotes. A plumber or an electrician sends a quote, gets busy, and three weeks later realises nobody chased it. A weekly nudge to the person who sent the quote, listing the ones with no reply after seven days, catches those before they go cold. The quote follow-up guide covers how to do that without pestering anyone. Job sheets that need copying from a job app into the accounts are another good candidate, as long as the app can export them.
For a finance team, it's the weekly checks: who's overdue and by how much, which supplier invoices are waiting for approval, whether the bank feed has stopped. None of those need a person to look, they need a person to act once something has looked for them. Where to start automating finance admin goes deeper on the order to tackle them.
For an office manager, it's the stuff that comes in by email and gets retyped. Web form enquiries copied into a spreadsheet. Holiday requests copied into a calendar. Supplier price lists saved into a folder and renamed. Those are small, frequent and visible when they go wrong, which is the test exactly.
Which jobs pass the test?
Here's a sample of common candidates run through it. Yours will differ, but the pattern usually holds.
| Candidate | How often | Systems involved | Good first automation? |
|---|---|---|---|
| Morning email listing today's jobs or bookings | Daily | Booking system or job app, Outlook | Yes |
| Overdue invoice reminders to customers | Weekly | Xero, QuickBooks or Sage | Switch on the built-in reminders first; build only if they fall short |
| List of unanswered quotes sent to whoever quoted | Weekly | Quotes spreadsheet or CRM, Outlook | Yes |
| Client appointment reminders | Daily | Salon or clinic booking system | Usually already built in; switch it on |
| Web enquiries copied into a tracking sheet | Daily | Website form, Excel or SharePoint | Yes |
| Weekly takings pulled into the accountant's spreadsheet | Weekly | Till or card system, Excel | Yes, if the till can export |
| Reply to a customer complaint | When it happens | Inbox | No, it needs a person |
| Year-end stocktake reconciliation | Once a year | Stock system, spreadsheets | No, too rare to repay the build |
| Commission calculation with changing rules | Monthly | CRM, payroll, spreadsheets | Not first; fix the rules before automating them |
What should you leave alone?
This is the part that saves money, so it's worth being blunt about.
Anything that needs judgement. Deciding whether to extend credit to a customer who's paid late twice, choosing which job gets the van when two clash, working out whether a refund is fair. An automation can gather the information for those decisions and put it in front of the right person. It shouldn't make the call.
Customer conversations that need warmth. A reminder that an invoice is overdue is fine to automate. The conversation with a long-standing customer who's clearly struggling is not. Same with complaints, condolences, and anything where the customer will notice a template. People can tell, and a cold automated reply at the wrong moment costs more than it saves.
Processes that change every month. If the way you do something keeps shifting, an automation will be out of date before it's paid for itself. Commission schemes that get tweaked each quarter, reports whose layout the board keeps changing, a pricing model that's still settling. Let the process settle first. Automating a moving target just means paying to rebuild it.
Things that happen twice a year. Year-end, the annual insurance renewal, the summer stocktake. They're painful, but they're rare, and by the time they come round again something will have changed. A good checklist does more for those than any automation.
There's a quieter fifth one too: anything where nobody can say what "right" looks like. If two people in the office would do the job differently, the automation will do it a third way and everyone will think it's broken. Agree how it should work, then automate it.
How do you start without it becoming a project?
Pick one job. Not a programme of work, not a roadmap, one job that passes the test. Write down what happens now: where the information comes from, who looks at it, who they tell, and what they tell them. That's the spec.
Then check whether your existing software already does it. If it does, you're finished. If it doesn't, build it over the systems you already use. For a business on Microsoft 365, that's usually Power Automate, which comes with most Microsoft 365 business plans for anything that only touches standard connections such as Outlook, Excel, SharePoint and Teams. The cost guide covers what it costs when it doesn't.
Run the automation alongside the manual way for a week or two. If the 7am email and the person who usually checks agree every morning, switch the manual check off. If they don't, you've found either a bug or a step nobody wrote down, and both are worth knowing about.
Once the first one is running, the second one is easier, because you trust it and you know what good looks like. That's how automation actually spreads through a small business: one boring job at a time, each one visible.
If you'd rather someone else picked the first one, my automation check is a 90-minute call that ends with a written list of up to three automations worth building, each with a fixed price, and it'll say so plainly when your software already does the job. It's £295, credited against a build.
Common questions
What is the easiest thing to automate in a small business?
A daily or weekly summary email is usually the easiest, because it only reads data and sends a message. It doesn't change anything in your systems, so a mistake is harmless and obvious. Overdue invoice reminders are just as easy if your accounts package has them built in, since that's a setting rather than a build.
Should a small business automate customer emails?
Routine ones, yes: booking confirmations, appointment reminders, polite overdue invoice nudges. Conversations that need judgement or warmth, no. Complaints, difficult payment conversations and anything personal should come from a person, because customers notice a template at the wrong moment.
Do I need new software to automate my business?
Usually not. Xero, QuickBooks and Sage have invoice reminders built in, booking systems send their own reminders, and Power Automate comes with most Microsoft 365 business plans for flows that use standard connections like Outlook, Excel and SharePoint. Start with what you already pay for and only add a tool when there's a gap.
How do I know if a process is worth automating?
Run it through four questions. Does it happen daily or weekly? Does someone currently check it by eye? Does the data already live in a system you pay for? Would a wrong answer be obvious quickly? If the answer to all four is yes, it's a strong candidate.
What shouldn't a small business automate?
Anything that needs a judgement call, customer conversations that need a human tone, processes that keep changing, and jobs that only happen once or twice a year. Those either cost more to maintain than they save or do damage when they get it wrong. Automate the gathering of information for those decisions, not the decisions themselves.
Is workflow automation worth it for a very small business?
It can be, even for a team of two or three, as long as the job is frequent. A ten-minute daily check is around forty hours a year, which is a week of someone's time. The trick is starting with one small job rather than trying to automate the whole business at once.