Insights · Automation · October 2026
The 12 tasks a small business should automate first
One 2026 survey found that three in four UK small business owners had already tried AI, and were still losing the equivalent of ten working weeks a year to admin. Trying it is not the problem. The list of what to automate is published everywhere and it is the easy half: the order you take it in is what decides whether any of it pays you back.
Where the week actually goes
Owners report working around 52 hours a week, with administrative work taking about 36 percent of it. Call that nineteen hours, two and a half days, spent before anything is sold or delivered. Financial admin on its own accounts for five to ten of those hours in most surveys, and invoice chasing is the task owners most often describe as the one they are permanently behind on.
The returns are smaller than the pitch decks suggest, and they are still worth having. Automating routine financial admin is reported to give back six to eight hours a week. Invoicing and payment reminders alone, in an Intuit-sponsored study, returned three to five. That is most of a working day, recovered from two tasks nobody enjoys.
Twelve tasks, ranked by what they leak
One rule sets the order. Automate where money is leaving the business first, where hours are leaking second, and where attention is going third, because an hour you save is worth something real and a job you never knew you lost is worth considerably more than that.
Where money leaks. Start here even if these are not the tasks that annoy you most.
1. Calls that go unanswered, picked up and qualified instead of going to voicemail. 2. Invoices issued on completion, with payment reminders that go out without you deciding to send them. 3. Quotes and proposals followed up on a schedule rather than when you remember. 4. Bookings and reschedules handled without the four-message exchange.
Where hours leak. These are the ones that make the week feel full.
5. Enquiries from the form, the inbox and WhatsApp landing in one place with one owner. 6. Appointment reminders and no-show follow-up. 7. Receipt capture and expense coding, which is the single most reliable return in the list. 8. The recurring report somebody rebuilds by hand every Monday morning.
Where attention leaks. Worth doing, worth doing last.
9. Review requests sent after a completed job, while the customer still remembers you. 10. Customer records updated from the conversation that just happened. 11. Reorder and stock triggers, where the rule is already in somebody's head. 12. A morning summary that tells the owner what happened yesterday and what needs a decision today.
Notice what is not on the list. Nothing here writes your marketing, sets your prices, or talks to a customer about anything that carries a consequence, and that is deliberate.
What to fix before you automate it
Industry analyses through 2026 put the share of automation projects failing to meet expectations near 70 percent, with around 68 percent of failures abandoned inside eighteen months. The most cited reason is blunt: roughly 73 percent fail because a broken process was automated rather than repaired. Weak ownership after go-live accounts for a further 35 percent.
So there are four things we will tell a client to leave alone for now.
Anything your team does differently each time. If two people handle lead follow-up in two different ways today, automating it produces inconsistency faster and at greater volume. Write the process down first. If you cannot write it down, that is the finding.
Anything you do fewer than about five times a week. The build cost will outrun the saving, and it will keep costing you in maintenance after the saving stops being interesting.
Anything where the first ninety seconds carries a legal, financial or clinical consequence. A person answers that. We do not negotiate on it.
Everything at once. Firms that automate eight things in a quarter end up with eight fragile systems and nobody who understands any of them, which is how you arrive at the eighteen-month abandonment number from the top of this section.
How to run the one you pick
Take the top item on your list that is already consistent, and do only that one.
Count it before you touch it. How many invoices went out late last month, how many calls rang out, how many quotes never got a second message. A number you wrote down in October is the only way to know in January whether the thing worked, and most automations that get abandoned were never measured at all.
Give it a number and a date. "Save time" is not a target and produces no failure signal. "Cut the average days to payment from 41 to 25 by the end of January" is a target, and it tells you by a known date whether to keep going.
Give it an owner. One named person who notices when it breaks. Systems without an owner drift and fail silently, which is worse than failing loudly, because you keep believing the work is being done.
Run the second one only when the first is boring. Boring means nobody checks it any more and it has not surprised anyone for a month.
The honest limit on all of this: automating your admin does not make a business grow. It gives back somewhere between three and eight hours a week, and what happens to those hours is entirely down to what you choose to do with them. Owners who use that time to call customers see something change. Owners who fill it with more admin have bought themselves an expensive way to stay exactly where they were.
Find the one that is leaking money in your business
We start every engagement by counting, not by building. Tell us what the week looks like and where the work piles up. We will come back with the two or three tasks worth automating in your business, in order, with the numbers they should move and the ones we think you should leave alone for now.