mmitech

Guide · · by Riki Baker

What admin really costs a UK small business — and what to automate first

Nobody invoices you for the paperwork, so it feels free. The studies say it's costing you about a month a year. Here's where the time actually goes, and the order I'd fix it in.

Nobody ever sends you a bill for admin, which is exactly why it feels free. The quote you write up at nine at night. The invoice you chase on a Sunday because that’s the only quiet hour you get. The phone that rings while you’re under a sink, goes unanswered, and turns into a job that quietly goes to someone else. It all feels like the cost of doing business. Most of it is the cost of doing business by hand, and it’s bigger than it looks precisely because it never shows up anywhere.

Sage put a figure on it that’s worth sitting with. Sage found that the average small business effectively works thirteen months to get paid for twelve, with roughly two days of every month lost to financial admin like raising and chasing invoices (Sage, 2025). A whole extra month a year, gone on work you can’t sell.

Owners say much the same about their own week. Ipsos found that 84% of small businesses reckon up to half their time goes on paperwork (Ipsos). And when the admin doesn’t get done on time, it costs you again: according to gov.uk, citing Intuit QuickBooks, late-paid invoices alone drain an estimated 56 million hours of lost productivity out of the UK economy every year (gov.uk).

You never feel any single piece of this. Four minutes copying an address from your inbox into the invoicing app doesn’t register. It’s the thousand four-minute jobs a year that quietly eat the month.

It helps to know what “admin” actually is

“Admin” is too vague to fight. Pull it apart and nearly all of it is one of four things.

There’s re-typing data you already have. The customer’s name and address are sitting in your inbox, your quote, your job sheet and your invoice, and you keyed them in four separate times. There’s chasing, whether that’s a quote gone quiet or an invoice drifting past its due date. There’s remembering: the services, the certificates, the “try me again in the spring” notes that live on a whiteboard or in your head. And there’s answering the same handful of questions over and over, usually while you’re holding a tool and can’t really get to the phone properly.

You didn’t train for any of it, and it’s certainly not why you started the business. It is, though, exactly the kind of thing software is genuinely good at. Not “AI runs your company”, just a small, reliable system doing the boring part the same way every time, so it stops landing on your Sunday.

Putting a number on your own week

Studies are useful, but they’re somebody else’s average. The number that lands is the one you work out for your own shop, and it’s not hard to do. Take a normal week, tot up the hours that go on the four things above, and put your real cost per hour against them. Not what you charge a customer — what an hour of your time is genuinely worth when there’s a queue of jobs behind it.

Say you run a two-person trades office. One of you is out on the tools most days; the other is half on site, half answering the phone and pushing paper. Here’s a rough week.

TaskHours/weekCost/year (@ £25/hr)
Re-keying job and customer details4£5,000
Chasing invoices3£3,750
Answering the same questions3£3,750
Total10£12,500

Illustrative figures.

Ten hours a week, at a blended twenty-five quid an hour, across fifty working weeks, is twelve and a half grand a year. Round it up for the bits you didn’t count — the context-switching, the jobs that slip while you’re heads-down on paperwork — and you’re knocking on thirteen thousand. That’s not a tools bill or a van you can see on the drive. It’s a salary’s worth of someone’s time, spent on work nobody pays you for, every single year.

The point of the exercise isn’t the exact figure. It’s that once you’ve written it down, “I’ll sort the admin one day” stops being a vague good intention and starts looking like the most expensive thing you’re putting off.

What to fix first

You don’t take all four on at once. Go for the one costing you actual money today, not just time.

Getting paid comes first, almost always, because it turns admin you hate into cash you’re already owed. A chase that goes out the moment an invoice tips past due, in a fixed order, polite then firmer, beats you remembering to do it on a Sunday and feeling awkward about it. That’s the whole idea behind an invoice and dunning ladder — the same nudges, every time, without you in the loop. I went deep on the why and the wording in a separate guide on automating the chase for late payments.

After cash, it’s not losing the enquiry. A missed call or an email left sitting for two days is a job you never even knew you had, and you can’t chase what you never logged. If the phone goes while you’re up a ladder, something needs to catch it, take the details, and put them somewhere you’ll actually look. I wrote up the cheaper ways to do that in never miss the enquiry.

Then the recurring work that can book itself: services, certificates, renewals, anything with a date attached to it. A gas safety check that’s due in eleven months is a job already on the books — it just needs something to remember it for you and send the reminder at the right time.

The fourth one, the repeat questions, is worth pulling forward if it’s eating real hours. “What are your opening times”, “do you cover my postcode”, “how much for a boiler service”, asked twenty times a week and answered twenty times badly because you’re mid-job. A trainable inbound AI agent can field exactly those, in your words, and only hand the awkward ones to you. It’s the difference between a phone that interrupts you all day and one that only rings when it matters.

Cash, then the work coming in, then the work coming back, then the questions. The re-typing and the filing matter too, but they’re rarely where you start, because they’re the cheapest to live with for another month.

Why most software doesn’t help

The big off-the-shelf tools assume your admin is generic. It isn’t. A letting agent’s chase has a legal order to it. A gas engineer’s reminder is tied to a CP12 date. A dental practice’s “recall” is really a treatment plan about to go cold. The admin is shaped like your business, so the thing that removes it has to be shaped like your business too, not like the average of ten thousand others.

There’s a quieter risk in the generic tools as well. Most small firms have one job that runs entirely on the one person who knows how it works — the bookings, the chase, the renewal dates, all living in their head or their notebook. That’s fine right up until they’re off sick or hand their notice in, and then nobody can find the thread. I’ve written about that exact failure mode in the one person who knows how it works. Off-the-shelf software doesn’t fix it, because it was never built around how your one person actually does the job.

That’s what we build: small, load-bearing apps that take a specific piece of manual admin off a specific business and make it disappear, scoped to how you actually work. If there’s one job in your week that runs on memory, sticky notes, or that one person, that’s the place to start. Send us a line about what it is, and we’ll come back with a sketch of what we’d build and roughly what it would cost — start here.

Tell us what's slowing the business down

Send an enquiry - what you do, what's slowing you down, what you've already tried. We'll come back with a sketch of what we'd build and what it would cost. No calendar, no demo to sit through.

No calendar widgets. Email reply, scoped sketch.

Tell us what's slowing the business down

Email reply, scoped sketch, you decide. No calendar widgets, no demo to sit through.

No calendar widgets. Email reply, scoped sketch.