Business Process Automation: A UK Owner's Guide to Automating the Handoffs

When the ONS asked UK firms why they were not using AI, the most common answer was not cost, and it was not skills. It was that they could not work out where to use it. I recognise that answer from almost every first conversation I have with an owner about automation.

I think the reason is the way business process automation gets explained. Most of what ranks for the term is written by IBM, Microsoft and Red Hat for companies with an IT department, so it talks about platforms and governance. What a ten person firm in Leeds needs to hear is simpler: your hours are not lost inside the tasks, they are lost in the gaps between them. That is what this guide is about.

What is business process automation?

Business process automation (BPA) is using software to run a repeatable process from the moment it starts to the moment it finishes, including every handoff between people and systems. The aim is that nobody spends their afternoon chasing what should happen next.

The key word is process. A process has a trigger (say, a supplier invoice landing in the accounts inbox), a series of steps, and an end point where the work is genuinely done. Automating one step inside it is useful. Automating the whole chain is a different and much bigger prize.

Is business process automation the same as RPA, BPM or AI automation?

No, though the terms overlap and vendors use them loosely. Here is how they differ in practice.

For a small business the useful takeaway is this: RPA is a workaround for software that will not talk to anything, and you should avoid it where a proper connection exists. BPA is the goal. AI is what now lets BPA reach the steps that used to need a human to read something.

What is the difference between automating a task and automating a process?

A task automation does one step faster. A process automation removes the waiting and the chasing between steps, which is usually where far more of the time goes.

Think about how a quote becomes paid work in a small trades firm. Sending the quote takes ten minutes. Then it sits. Someone checks whether it has been accepted, someone else books the job, someone creates a deposit invoice in Xero, and the customer rings to ask when you are coming. Each step is quick. The process takes nine days, and most of those days are people waiting on other people.

A tool that writes quotes faster saves you ten minutes. An automation that notices the acceptance, books the slot, raises the invoice and confirms the date to the customer saves you the nine days, and the phone call.

Most small firms I meet have automated a few tasks. Very few have automated a process. That gap is the opportunity.

Where do the hours actually hide in a business process?

They hide in four places, and none of them look like work while you are doing them.

1. Retyping. The same client name and reference typed into HubSpot, then again into Xero, then once more into a spreadsheet.
2. Waiting. A step sits in someone’s inbox until they notice it, often for days.
3. Chasing. Emails and calls whose only purpose is to find out whether the previous step happened.
4. Status questions. “Where are we with the Harrison job?” asked by a customer, answered by digging through three systems.

The American Express SME Barometer, published in July 2026 from a survey of 1,000 UK small business owners, found owners spend around 11 hours a week on admin and finance. That is roughly twice the time they spend on sales. My strong suspicion is that most of those hours are in the list above, not in any single task you could point to.

What does business process automation look like in a real process?

Take client onboarding at a small accountancy practice, because it has everything: a clear trigger, regulated steps, several systems and plenty of waiting. Here is how I would lay it out.

Notice what did not change. The money laundering check still happens, and a qualified person still reviews it. The HMRC authorisation still needs the client to act. The automation does not cut corners on the regulated parts. It just stops those parts from sitting in someone’s inbox for a week.

Notice too that no single step in the “after” column is clever. Pulling a company number from Companies House is a few lines of code. The value is entirely in joining the steps up so the whole process runs from one signature.

Which parts of a process should stay with a person?

Keep a person on any step where a mistake would move money or reach a regulator. Automate everything that leads up to that step, so the person arrives with the information already gathered.

In practice that means the automation does the collecting and the preparing, then stops and asks. An approvals card in Microsoft Teams works well, and so does a plain email that says “this one needs you”. The mistake I see most often is the opposite: owners automating the judgement and keeping the copying.

AI has moved this line, but less than the marketing suggests. A model like Claude can now read a supplier invoice or sort an enquiry into the right category very reliably. I would still let it prepare the decision rather than make it, at least until you have watched it get a few hundred right.

Why does business process automation fail in small firms?

Usually because nobody owns the whole process. The accounts person built one Zap, the office manager built another, and when a step breaks in between, nobody notices until a client complains.

The second reason is that the process was never written down, so the automation copies the version in one person’s head, exceptions and all. Think of the supplier who always sends invoices as photos taken on a phone. That is exactly where automations quietly fall over.

The fix for both is the same, and it is the part I care about most. The person who owns a process should be the one who can change its automation. For years that was unrealistic because it meant learning to code. It is not unrealistic now. With Claude Code, the office manager who knows the onboarding process better than anyone can describe it in plain English, have Claude write and test the code, and change it themselves when HMRC changes a form. You own the skill instead of renting an agency every time something moves.

You also probably own more tooling than you think. Most Microsoft 365 business plans already include Power Automate cloud flows with the standard connectors, at no extra cost. Many firms pay for Zapier while an unused automation tool sits in the subscription they already have.

How do you choose which process to automate first?

Choose the process with the most handoffs. Count how many times the work passes from one person or system to the next. Four or more is a strong candidate. Then check it starts with a clear trigger you could point at, like a form submission, a signed document or an email arriving in a particular inbox.

If you want a quick read on where the hours in your business are going before you pick, the AI Opportunity Scorecard takes about two minutes and is free.

Where I would start

If this were my business, I would pick the one process that makes customers ring up to ask where things are. Those calls are the clearest sign that work is stuck between steps, and fixing that process pays twice: once in your team’s hours, and again in how organised you look.

I would map it on one page, automate the handoffs first, and leave the judgement with a person. If you would like help doing that on a real process of yours, the AI Opportunity Audit is a £500 fixed fee, 60 minute session where I review how your business runs, pick the highest impact place to start, and build a working automation live in Claude Code while you watch. You leave with a prioritised roadmap and the confidence to build the next one yourself. You can read more about how I work on the about page.