Most businesses come to us with a vague sense that they’re spending too much time on things a computer should be doing. What they don’t always have is a clear picture of which things, and in what order to tackle them.
This post walks through how we approach that question.
Key Takeaways
- The best automation candidates are rule-based tasks that repeat on a schedule or trigger — not tasks requiring genuine judgement
- The three patterns that account for most manual work: follow-up sequences, data moving between systems, and outbound communication that follows a template
- Rank by cost of failure, not by what seems easiest to automate
- Most automations are live and working within a few weeks
What is business automation?
Business automation means using software to handle repetitive, rule-based tasks that previously required a person to do manually. It isn’t about replacing people — it’s about freeing them from the tasks where a computer can apply the same rule faster and more reliably than a human can.
The key test: if you can describe the process as a set of conditions and outcomes — if this happens, do that — you can automate it. If the process requires context, intuition, or judgement that changes each time, you can’t (yet).
For UK SMEs, the most common starting points are sales follow-up, client onboarding, invoice chasing, CRM data entry, and outbound reporting. These aren’t glamorous. But they’re where the time goes.
Start with your calendar, not your tools
The most reliable way to find automatable work isn’t to audit your software stack — it’s to look at your last two weeks of calendar and ask: which of these recurring tasks required a human to make a real decision?
Anything that didn’t — follow-up emails sent on a schedule, reports pulled and formatted, invoices chased after 30 days — is a candidate for AI automation.
The distinction matters because automation is good at executing rules, not at exercising judgement. Once you’ve defined the rule clearly enough to write it down, you’ve usually defined it clearly enough to automate it.
The three patterns we see most often
Across the UK businesses we’ve worked with, the manual work that shows up most reliably falls into three categories.
1. Follow-up sequences that depend on time or status
Sales leads that go cold because nobody chased them. Invoices that go unpaid because the reminder wasn’t sent. Clients who fall off the onboarding process because the next step wasn’t triggered.
These are rule-based by nature. The business already knows what should happen — it just isn’t happening automatically.
Recruiter example: New candidate registers → automated confirmation and next-steps email → three days later, if no response, follow-up sent → status updates in the ATS without anyone touching it.
Accountant example: Client submits documents → acknowledgement sent → assigned to the correct team member → reminder triggered at day seven if outstanding work hasn’t started.
Agency example: Project marked complete → automated client satisfaction email → if positive, review request sent → response data logged in CRM.
2. Data that lives in the wrong place
Information collected in one system that needs to appear in another. New enquiries that should create CRM records. Completed jobs that should trigger invoices. Reports that should consolidate numbers from three different dashboards.
Moving data between systems is exactly the kind of work AI automation handles well. It’s deterministic, repetitive, and the cost of not doing it — stale CRM records, missed invoices, manually assembled spreadsheets — is easy to measure.
3. Outbound communication that follows a template
Review requests, renewal reminders, project status updates, monthly reporting. These are high-value touchpoints that often don’t happen at all because someone has to write and send them manually.
Automation doesn’t make them impersonal. It makes them actually get sent — every time, on schedule, to every client.
What makes a good automation candidate?
Not every process is worth automating. Before committing to a build, check the process against this list:
- It repeats — the same steps happen more than a handful of times a month
- It follows a rule — the outcome is determined by conditions, not judgment calls
- The cost of failure is real — missing it loses money, causes friction, or damages trust
- The inputs are consistent — the data coming in is structured enough to act on programmatically
- Someone currently owns it — if no one is doing it at all, automation won’t fix the underlying problem
If a process clears all five, it’s ready to scope.
The question that tells you where to start
For each process you identify, ask: what’s the cost of this not happening?
A lead follow-up that doesn’t go out has a measurable revenue cost. A report that doesn’t get sent causes friction and trust erosion. An invoice that isn’t chased directly hits cashflow.
Rank by impact, not by what seems easiest to automate. The highest-value processes are usually not the most complex — they’re just the ones nobody has got around to fixing yet.
What happens after you identify the work
A scoping conversation is the right next step. We look at the specific process, the systems involved, and what a working automation would actually need to do. From there we can give you a clear scope and a project price.
Most automations we build are live and working within a few weeks. The processes that were consuming hours a week keep running — they just stop requiring anyone’s attention.
If you want to see what’s possible across different types of automation work, the what we build section covers how we approach each one.
If you want to talk through what’s worth automating in your business, book a call.
Frequently asked questions
What is business automation?
Business automation means using software to handle repetitive, rule-based tasks that previously required a person to do manually — things like sending follow-up emails, moving data between systems, chasing invoices, or generating reports. The process still happens; it just stops requiring someone’s time.
How long does it take to build an automation?
Most automations we build are live and working within two to four weeks. The timeline depends on how many systems are involved and how clearly the process is already defined. If you can describe the rules, we can usually build it quickly.
How do I know if a process is worth automating?
Ask what it costs when it doesn’t happen. A lead follow-up that gets missed has a revenue cost. An invoice that isn’t chased hits cashflow. A report that doesn’t go out causes friction. If the cost of failure is real, the process is almost certainly worth automating.
Can AI automation work for a small business?
Yes — and small businesses often benefit more than larger ones, because the time cost of manual work falls on a smaller team. You don’t need a large budget or an in-house tech team. You need clearly defined processes and the willingness to invest in fixing them once.
What’s the first step to automating our business?
Map your last two weeks of recurring tasks and ask which ones required a human to make a genuine decision. Everything that didn’t is a candidate. Then rank by impact — start with the processes where failure has a direct cost — and book a call to talk through the specifics.