Automation has become a big word. A flowchart on the screen, boxes joined by arrows, people talking about integrations and triggers. A small business owner looks at that and concludes, quite reasonably, that they will never finish.
I started making progress when I stopped asking "what can I automate" and started asking something else: what currently depends on my memory and shouldn't?
The answer to that question fits in a short list, in order. And the order matters more than the list, because almost everyone starts with the fun part and quits before reaching the part that makes money.
Automating isn't building a robot, it's no longer depending on your memory
Forget the flow diagram. There is only one measure: return per unit of effort.
On one side, what it costs to set up, in your time and in the risk of getting stuck halfway. On the other, what it gives back every week, forever, without you remembering anything.
With that measure, things that look sophisticated drop to the back of the queue, and things that look almost silly move to the top. A saved reply in your messaging app that you use twenty times a week is worth more than any elegant integration that would take you three Saturdays to configure.
The goal isn't an impressive system. It's for the business to stop depending on you remembering to do something at the right moment.
Priority 1: responding to the lead
The enquiry comes in. A form filled out, a message, someone asking for a quote. That is the hottest moment in the entire relationship, and it's where most small businesses lose more money than in any badly built campaign.
It isn't just my impression: Harvard Business Review audited thousands of companies and concluded, in The Short Life of Online Sales Leads, that most of them don't respond anywhere near fast enough. The practical rule, and I watched it work exactly this way in my own business: a lead that waits goes cold. The person filled that form in with the problem hot in their head. A few hours later, either they solved it another way, or a competitor answered first.
Here, automating means two things, neither of them complicated.
The enquiry has to reach you immediately. The form on your page has to send the lead straight to your email and your phone, the instant it's submitted. Plenty of pages have a form that files the contact in a dashboard nobody opens. That isn't lead capture, it's a dead archive.
The reply has to already be written. You shouldn't be drafting from scratch at nine at night. Keep three or four templates ready for the cases that repeat, and personalise two lines. AI writes those templates in minutes, in your voice.
The speed is still yours. What automation does is remove every bit of friction between the enquiry landing and you answering it.
Priority 2: the three-email sequence nobody sends
Someone who left their details and didn't buy today is not a lost lead. In most cases, it's a lead still deciding.
Three messages solve nearly all of it:
- Thank you, the same minute. Confirms it arrived, says what happens next.
- Proof, a few days later. A real case, a photo, a testimonial, something that answers "does this actually work?".
- Offer, a few days after that. The direct invitation, with a clear next step.
You write it once. It runs forever, for every new contact.
That is the highest-return automation on the entire list, and it's the one fewest people build. Not because it's hard. Because it's invisible: nobody applauds an email sequence. It only shows up in the sales that close three weeks after first contact.
An email tool on a free plan handles this at a small business's volume.
Priority 3: quick replies and a catalogue in your business messaging app
Messaging is where a lot of the selling happens. And the business version of the app you already use gives you three things free that save time every single day.
Quick replies. Those ten questions you answer every week become shortcuts. You type two letters and the full answer appears. That isn't convenience, it's the difference between answering in a minute and answering whenever you get around to it.
A catalogue. The customer sees what you sell without leaving the conversation. It cuts the worst bottleneck in customer service, which is sending photos and prices item by item.
Profile and labels. A business profile with a logo and opening hours builds trust. Labels organise who's a new lead, who's negotiating and who's closed.
Use a business number, separate from your personal one. That lets you put someone on your team on it later, and keeps your life outside the business.
This priority takes under an hour and gives time back the same week.
Priority 4: a watchdog that tells you when the site goes down
This one is cheap, fast and almost never remembered.
A free monitor keeps an eye on your site and your campaign page every few minutes. If either goes down, you get an email within minutes.
Why it matters: silent failure is the most expensive way to break. A site can sit offline for days while you pay for ads sending people to a page that won't load. The watchdog turns "I found out because a customer complained" into "I fixed it in ten minutes".
Ten minutes to set up, zero maintenance afterwards.
Priority 5: content in batches, one session a month
Content can't be automated. It can be grouped.
Instead of deciding every day what to post, you do one session a month: you sit down with the AI, it builds the calendar and the copy for the whole month with your business context, you review, adjust anything that doesn't sound like you, and schedule the lot in one go. New blog articles come out of the same session.
What this kills isn't the work of producing. It's the daily decision, which is what actually wears you down.
It's last on the list on purpose. Content builds the medium term. The four items before it move this week's cash.
What not to automate right now, and why
Here's the part almost no article on automation writes, because it doesn't sell software.
I deliberately did not automate three things, and the business doesn't miss any of them.
API integrations between systems. The idea of the form talking by itself to a management system, which talks to the calendar, which talks to billing, is beautiful and it's a trap for anyone without technical support. Each of those connections is a point that breaks silently, on a day when you're busy, and you only find out when a customer chases you.
Webhooks and technical notifications. Same logic. It's the category of thing that works perfectly until it stops working without telling you.
Automated booking. Customers booking their own slot sounds like the dream, and for some businesses it is. For mine, the conversation is where the sale happens: that's where I understand what the family wants and offer the right thing. Automating that would have made me sell less.
The measure is the same as always: if something breaks and you have nobody to call, the cost isn't the time to build it. It's the business sitting still while you try to work out what broke.
This isn't laziness or a limitation of the tools. It's a choice of scope. There is a level above this, with code on a server and real maintenance. It only makes sense once the basics are running and turning a profit.
The time maths: what each item gives back
I'm not going to hand you a table of hours saved, because it would be made up. But the nature of the return is different for each item, and that's worth understanding.
Responding to the lead gives back money, not time. You won't work less. You'll close more of what already comes in.
The email sequence gives back sales that simply don't happen today. Zero ongoing effort once it's built.
The business messaging app gives back time directly, every day, in proportion to how much customer contact you handle.
The watchdog gives back losses avoided. You don't feel it until the day it warns you.
Batched content gives back mental load. It's the item that changes how you feel about marketing the most, and shows up least in the month's numbers.
Notice: only one of them obviously saves hours. The other four plug leaks. That's why the measure is return per effort, not "how much time do I save".
One detail you may have missed: priority 1 and priority 2 both depend on you having a page of your own, with a form that genuinely works and sends the enquiry to you. Without that, there's nothing to automate, because no leads are coming in in an organised way.
And before automating anything, it's worth having the number that tells you whether the automation paid off: what a new customer costs you today. The free Cost Per Customer X-Ray is a prompt you paste into a free AI, in the browser, with nothing to install. It asks a few questions about your spend and your new customers and gives you the figure in about five minutes. That's your "before". Without it, you'll never know whether the "after" moved.
Read next
- How much time a week your business's marketing really takes, to see whether the routine fits your diary.
- How to give the AI your business context before asking for anything, which is what makes those reply templates sound like you.