How Much Should a UK Small Business Budget for Automation?

Most UK small businesses with 5-40 staff should budget somewhere between £1,500 and £15,000 for a first automation project, depending on how many systems need connecting and how messy the underlying data is. That's the honest range I quote after eight years doing this work, not a headline figure designed to get you on a call. The real question people are asking when they type "how much does business automation cost for a small business" into Google is whether it's worth the money at all, and that depends entirely on what a task currently costs you in staff time. So before you budget anything, you need to know what you're actually paying for the manual version of the work you want to remove.

How much does business automation cost for a small business?

I've quoted projects from £800 to £22,000 over the past three years. The spread isn't random. A single workflow, such as pulling new enquiries from a web form into Xero as draft invoices, sits at the £800 to £2,500 end. A multi-system project, say connecting Reapit, Xero and a mailbox for a lettings agency with approval steps and error handling, runs £6,000 to £15,000. Anything above that usually means you're building something closer to bespoke software, and at that point I tell clients to talk to a proper development shop, not an automation consultant.

The variable that moves price more than anything else is data quality. Clean, consistent inputs (same file format every time, same field names, no manual re-keying halfway through) keep a project cheap. Messy data, three different spreadsheet formats depending on who filled it in, or a legacy system with no proper API, adds weeks and pounds. I've had projects double in scope purely because the client's product codes in Sage didn't match the codes in their e-commerce platform.

What does a typical automation project actually include?

A project quote should cover four things, and if a supplier's quote only covers one, ask why.

First is discovery: mapping the current process, usually a half-day to two days on site or on video calls. Second is build: the actual connections and logic. Third is testing with real data, not sample data, because sample data never breaks and real data always does. Fourth is a handover document and a short training session so your team can make small changes without calling me every time a field name shifts.

What a quote should never include, in my view, is an open-ended "support retainer" sold as mandatory. Some automations need ongoing monitoring because the systems they touch change often (anything involving HMRC-facing software falls into this category). Most don't. I build things to run unattended and only charge for maintenance when something in the underlying systems actually changes.

How do I work out if automation pays for itself?

This is the calculation that matters more than any quote you'll receive. You need the fully-loaded cost of the person doing the task now, not their gross salary, because salary alone understates the true cost by 25% to 40% once you add employer's National Insurance, pension contributions, holiday cover, and overheads.

Here's a worked example from an actual engagement, figures rounded for clarity.

An office administrator on a gross salary of £26,000 works 37.5 hours a week across 46 working weeks a year (52 weeks minus 5.6 weeks statutory holiday, rounded down).

• Annual working hours = 37.5 × 46 = 1,725 hours
• Gross hourly rate = £26,000 ÷ 1,725 = £15.07
• Fully-loaded hourly rate (× 1.3) = £15.07 × 1.3 = £19.59

That administrator spends 45 minutes a day manually re-entering supplier invoices from email attachments into Sage. Automating the extraction and entry cuts that to 5 minutes a day of checking, saving 40 minutes daily.

• Weekly time saved = 40 minutes × 5 days = 200 minutes = 3.33 hours
• Annual time saved = 3.33 hours × 46 weeks = 153 hours
• Annual value of time saved = 153 × £19.59 = £2,997

Against a build cost of £2,200 for that specific workflow, the project pays for itself in just under nine months and then keeps paying every year after. That's the calculation I run before I quote anything, and I'd encourage you to run it yourself before you call anyone, including me.

When should I use off-the-shelf tools like Zapier, Make or n8n instead of a custom build?

Honestly, more often than most consultants will admit. If your workflow is a single trigger and a single action, new row in a Google Sheet triggers a Slack message, there is no reason to pay someone £2,000 to build that when a £20-a-month Zapier plan does it in twenty minutes. I've turned down work because a client's need was fully solvable with a free Make plan and I said so on the discovery call.

Where these tools start costing you more than they save is volume and complexity. Once you're processing hundreds of records a day, or the logic has more than three or four conditional branches, per-task pricing on these platforms gets expensive fast, and debugging a tangled Zapier setup built by three different people over two years is a job I charge more for than building the same thing properly from scratch. n8n sits in a middle ground, since you can self-host it and avoid per-task fees, but then you've taken on hosting and updates as a technical responsibility, which is fine if someone on your team is comfortable with that and a liability if nobody is.

My rule of thumb: if you can sketch the whole workflow on one side of A4 with fewer than five steps, try the off-the-shelf tool yourself first. If it needs a diagram with branches and exceptions, get a proper quote.

When should I not automate at all?

I turn down roughly one in six enquiries, and it's usually one of two situations.

The first is when the process itself is broken, not just manual. If your quote-to-invoice process involves five people, three of whom don't know what the other two are doing, automating the current mess just makes the mess run faster and produces errors nobody notices until month-end. I had a client last year who wanted their stock reconciliation automated between two warehouses. The reconciliation process itself hadn't been agreed between the two site managers for over a year. I told them to fix the process on paper first and come back once there was one agreed method. They did, six weeks later, and the automation took three days to build once the logic was actually consistent.

The second is low-frequency, low-cost tasks. If a task takes fifteen minutes, happens twice a month, and the person doing it isn't a bottleneck for anything else, the maths rarely works. Fifteen minutes twice a month at a loaded rate of £20 an hour is £10 a month, £120 a year. No automation project pays for itself against £120 a year, and I'd rather tell you that on the phone than take a £1,500 deposit for it.

What UK systems do these automations usually connect to?

Most of my work touches a fairly predictable set of tools. On the finance side, that's Xero and Sage, pulling data in from bank feeds, e-commerce platforms or supplier emails. On the property side, Reapit and Alto come up constantly for lettings and estate agency clients, usually connecting tenant enquiries or maintenance requests through to email and calendar systems. Companies House data shows up in due diligence and onboarding workflows, particularly for accountants and legal practices checking new client details automatically rather than manually searching the register.

The common thread is that none of these systems were built to talk to each other by default. The value isn't in any single tool, it's in the glue between them, and that glue is what you're actually paying for when you commission an automation project.

Frequently asked questions

How long does a typical project take?

A single-workflow project usually takes one to two weeks from signed quote to handover. A multi-system project with several conditional steps typically runs four to eight weeks, mostly because testing against real, messy data takes longer than the build itself.

Do I need to hire a developer to maintain this?

Usually not. Well-built automations run without intervention for months. What you do need is one person internally who understands roughly how it works, so that when a system update changes a field name, they know who to call rather than assuming the whole thing is broken.

What's the cheapest way to start?

Pick your single most repetitive, most hated manual task, the one everyone complains about in the same meeting every month, and try to solve it yourself with a free trial of Zapier or Make before spending anything. If you can't get it working in a couple of hours, that's useful information, and it tells whoever you hire exactly where the real complexity sits.

Will this work with my accounting software?

Xero and Sage both have reasonably open APIs and I've built against both without major issues. Older desktop-only accounting software, or heavily customised legacy systems, is a different conversation, and I'll usually ask for API documentation before quoting anything involving a system I haven't touched before.

Where to start

This week, without hiring anyone, pick one manual task and run the arithmetic I showed above using your own numbers: gross salary, hours worked, minutes spent on the task daily or weekly. Multiply by 1.25 to 1.4 to get a fully-loaded hourly rate, then work out the annual cost of that task in its current manual form. If the annual figure is comfortably above £1,500, it's worth a proper conversation. If it's below £500, put your energy into something else this quarter.

If you want a second pair of eyes on that number before you spend anything, book a 20-minute call and bring the figures with you.