If what you're doing is repetitive, there is a better way

The line item you already pay for

In most of the businesses we work with, the biggest cost isn't rent, stock, or software subscriptions. It's people.

That's not a problem — it's the point. You hire capable people because the work needs judgement: the awkward customer, the supplier who's being difficult, the margin call on a quote, the decision about which order to bump when production is tight. That's where salaries earn their keep.

The problem is what those same people spend their days actually doing.

Watch a week of operations, finance, or customer services and you'll often find a large slice of the day isn't judgement at all. It's moving data:

• Copying a PO from an email into the order system
• Re-keying invoice lines from a PDF into Xero
• Pulling a customer record out of the CRM to answer "where's my order?"
• Exporting a spreadsheet from one tool, tidying it, importing it into another
• Chasing the same status update across three tabs because nothing talks to anything else

Nobody hired them for that. They're doing it because the systems don't, and because "that's just how we do it" has been true for years.

The sentence we keep coming back to

If what you're doing is repetitive, there is a better way.

Not every repetition is worth automating on day one. But if the same person does the same sequence of clicks fifty times a week, and the rules are mostly predictable, you're paying skilled wages for unskilled labour. The business feels busy. The people feel tired. Nothing strategic moves forward.

The better way is usually boring: data flows where it should, overnight jobs reconcile what needs reconciling, exceptions land in front of a human with the work already done. We've written about specific shapes of this — documents turned into structured orders, draft replies in customer services, quotes assembled from your own pricing — but the principle is older than AI. Automate what can be automated. Leave people on the bits that need a person.

What "needs a person" actually means

The intelligent work doesn't disappear. It changes shape.

A customer-services team stops spending twenty minutes per email looking things up and spends it on the angry one, the unusual request, the customer who needs a phone call. An order desk stops typing POs and starts confirming matches, handling pricing disputes, calling the customer when something doesn't add up. A finance team stops rebuilding the same report every Monday and starts on the conversation with the director about why margin slipped on that product line.

Those jobs are harder to hire for, not easier. You want the person who can sense when a customer is about to leave, not the person who can type fastest.

Automation doesn't replace that judgement. It protects the time for it.

It's usually not about cutting headcount

This is the conversation we want to have honestly, because the fear gets in the way of sensible projects.

Most owners aren't asking us to help them make five people redundant. They're asking us to help them not hire another five when the business grows another 20%.

The maths is blunt. If you're paying £35k–£45k all-in per head in operations or admin, and three people are spending half their week on repetitive data movement, that's roughly £50k–£70k a year of salary spent on work a machine should do. Not wasted in the sense that people are lazy — wasted in the sense that you're buying the wrong output from expensive inputs.

A well-scoped integration or workflow build often costs less than a year of that tax. Payback inside twelve months is common; inside a quarter isn't unusual when the volume is high.

And when you grow:

• Without automation — every 10% more orders means another body in the order desk, another person copying data, another inbox drowning
• With automation — the same team absorbs more volume; new hires go on growth, service quality, or new product lines — not on being the human API between systems

That's a much better problem. You're not shrinking the team. You're stopping the team from swelling every time revenue ticks up.

How to spot the repetitive work

You don't need a time-and-motion study. Ask these on a normal week:

1. What does this person do when they're "busy" but not on the phone with anyone? If the answer is mostly screens and typing, dig deeper.
2. Where do the same errors keep appearing? Re-keying errors are a tell — wrong product code, wrong address, wrong price tier.
3. Who is the person everyone goes to because they "know how to get the data out"? That's a human integration layer. It shouldn't need to be a career.
4. What would break if this person were off sick for a fortnight? If the answer is "nobody knows how to run the export", you've found a process, not a person.
5. If you doubled volume next quarter, what would you hire for first? If it's "someone to help with orders / invoices / emails", you're about to buy more repetition.

The discovery work we do is largely this: measure where time goes, map which systems should be talking, quote the cost of fixing it properly. The numbers are usually more obvious than anyone expects.

What good automation looks like (and what it doesn't)

Good:

• Runs in the background; nobody thinks about it until it fails
• Surfaces exceptions to a person with context already assembled
• Writes to the system of record — CRM, ERP, accounts — not a shadow spreadsheet
• Can be audited: what ran, when, what it changed
• Has a human override when the rules don't fit

Not good:

• A fancier spreadsheet that still needs manual babysitting every day
• A chatbot that answers customers without access to real order data
• A one-off script only the developer understands and nobody can run when they're on holiday
• Anything that removes humans from decisions that carry liability, tone, or relationship weight

The goal isn't lights-out unmanned operation. It's lights-on for the work that deserves a brain.

Where to start

You don't automate the whole business at once. You pick the highest-volume repetition with the clearest rules:

One workflow, wired properly, often frees more time than people expect — because the repetition was fragmented across six people doing slightly different versions of the same bridge.

A first useful version is often four to eight weeks. The question isn't "can we afford to build it?" It's "can we afford another year of skilled people acting as the integration layer between tools that should have been talking years ago?"

The point, plainly

Your staff are the biggest cost and the biggest asset. Spending their time moving data around isn't a good use of either.

If what they're doing is repetitive, there is a better way — and it's usually not about having fewer people. It's about not needing more of them just to keep the plates spinning.

If you'd like to walk through where the repetition lives in your operation and what a sensible first automation might look like, say hello.