Skip to contact
Seamwork AI
← Blog

When does AI break even?

For a small business, AI needs to save somewhere between about one hour and four and a half hours a week in its first year to pay for itself, and which end you're at depends almost entirely on whose time it gives back. If it frees up hours you can sell, it pays back easily. If it frees up hours you'd have paid for anyway, it often doesn't.

Here's how that sum works, so you can put your own numbers in.

Where the "5.2 hours a week" figure comes from

The number you'll hear most is 5.2 hours a week. It comes from research by OpenAI, reported by Enterprise Nation in May 2026, based on a survey of 1,000 UK SME decision-makers.

It's worth knowing two things about it. The research was paid for by a company that sells AI. And it's people estimating their own time savings, not anyone measuring them. That isn't dishonest. It's simply how a survey works: someone asks a thousand business owners how much time AI saves them, and on average they say a bit over half a working day. But a feeling about time saved isn't the same as a measurement of it.

When someone measured it properly, the result was different

In July 2025, the research group METR ran a randomised controlled trial, the same kind of fair test used for new medicines, where tasks are split at random so the only real difference is the thing being tested. Sixteen experienced software developers believed afterwards that AI had made them 20% faster. When the clock was checked, they'd actually taken 19% longer.

METR are careful about what this shows, and we should be too. It's software development, sixteen people, and one snapshot of early-2025 tools, and the authors say plainly that it doesn't prove AI fails to speed up most developers. What it does show is a wide gap between how fast people felt and how fast they were. That's a good reason to treat 5.2 self-reported hours as the most you might hope for, rather than what to expect.

So for the sums below, we've halved it. That's our judgement, not a finding: call it two and a half hours a week for one person, or roughly 120 hours a year once holidays are taken out.

An hour is only worth money if something changes

This is where most payback sums go wrong. In a small business there are three kinds of hour, and they're worth very different amounts.

An hour you could bill. It's worth whatever you charge, but only if you'd otherwise be turning work away. If your diary is already full, an hour given back is an hour sold. If it isn't, it's an earlier finish, which is worth having but doesn't show up in the accounts.

An hour of admin time. From April 2026 the National Living Wage is £12.71 an hour for those aged 21 and over. Employer's National Insurance at 15% on earnings above £5,000 a year, plus the minimum pension contribution, pushes the true cost a pound or two higher. On the other hand, a business your size may find the £10,500 Employment Allowance covers its National Insurance bill entirely. The allowance is a discount the government gives employers off what they owe. Taken together, call it £13 to £15 an hour.

An hour nobody was ever going to bill. It's worth nothing in cash. It still matters, but it belongs under "less stress", not under "return".

For comparison, median hourly earnings excluding overtime for UK full-time employees were £19.67 in April 2025, according to the ONS. Most of the hours AI can realistically give back are cheaper than that, because the jobs it does best tend to be the routine ones.

What it costs

The costs are the easier half, because most of them are published.

Subscriptions. The Microsoft 365 Copilot Business add-on is £13.80 per user per month on an annual commitment. That's a promotional price running to the end of December 2026, against a standard £16.10, or £19.32 if you pay monthly, all before VAT. Claude Team is $20 per seat per month billed annually, or $25 monthly. For three people, budget somewhere between £500 and £700 a year.

Setting it up. We should be upfront here: we sell this, so we're not a neutral party. Our work starts at £2,500, paid half to book and half once it works. Anything above that is quoted after a free call, against what the job is worth to you. Other people charge other amounts, so use whichever figure applies to you.

At our starting figure, a three-person business is looking at roughly £3,100 in the first year. After that, it's the subscriptions, around £600 a year, plus whatever you agree for looking after it once it's running, which we quote rather than publish.

If you'd like to talk through what setting up AI for one of your own jobs would involve before you do the sums, we're happy to have that free call with you.

The break-even sum

Break-even is simply the point where what you've saved matches what you've spent. To find it in hours a week, divide the yearly cost by 48 working weeks, then divide by what one hour is worth to you. That's the whole method.

At £60 an hour, a plausible billable rate for a small professional firm (use your own), the first year needs about 1.1 hours a week saved to break even. After that, on the subscriptions alone, it needs about twelve minutes. Even a heavily discounted version of the headline figure clears that comfortably. This is the case where the answer is a straightforward yes.

At £14 an hour, the first year needs about 4.6 hours a week, and the subscriptions alone need just under an hour a week after that. Four and a half hours is nearly double our halved figure, and close to the full, self-reported 5.2. You'd be relying on the survey being right.

That comparison is the heart of it. Same tools, same time saved, and the sum either works twice over or doesn't work at all, depending on whose hours you're freeing up.

When it probably won't pay back

When the time saved belongs to lower-paid staff who stay on the same hours. Research by the British Chambers of Commerce and Atos, analysed with the University of Essex, found that 95% of SMEs using AI reported no impact on workforce size over the past year, and 86% said job roles were unchanged. The hours saved may well have been real. They just didn't turn into money, because nobody's wage bill moved and nothing extra was sold. If the freed time won't reduce paid hours or bring in more work, the saving is comfort rather than return. That's still worth something, but it's worth being clear-eyed about.

When the work is varied and doesn't come round often. A large experiment with Boston Consulting Group consultants found that people using AI got faster and better on tasks it handles well, and less likely to reach the right answer on a task chosen to sit just outside what it could do. Same tool, same people, opposite result. Repetitive, frequent, similar-looking work is where the gains are, and it's also the only kind you can measure well enough to know afterwards.

When the person doing the work is already very good at it. In a study of customer support agents given an AI assistant, most of the gain went to newer and less-skilled staff, with very little change for the most experienced. If the hours you want back belong to someone who's done the job for fifteen years, expect the smaller end.

What we'd suggest doing next

Pick one job: the one that comes round most often and looks most alike each time. Count it honestly for a fortnight. How many times did it happen, how many minutes did it take, and who did it? Multiply the minutes by what that person's hour really costs you, and compare it with the yearly cost of the tools plus the setup.

If the number doesn't clear the bar, it's probably not worth doing yet, and that's a perfectly sensible place to land. A business of three shouldn't buy AI on the strength of a survey, and a fortnight of counting costs nothing.

If the sum does work, it'll usually be because your own hours are the ones being freed and you have somewhere to sell them. If that's where you've got to and you'd rather not set it up yourself, we're happy to do it for you.

Sources

Checked on 29 October 2026. Software prices move quickly and the Copilot figure above is explicitly promotional until the end of December 2026 - check both before you rely on them.

The next step

The cost of waiting is not zero.

Somewhere in your market, a business the same size as yours has stopped writing follow-ups by hand. They are quoting faster than you, chasing money you are still chasing manually, and starting Monday already knowing where they stand. That gap does not stay the same size.

Half an hour on the phone. You describe your week, we tell you honestly which parts of it are worth automating and which are cheaper left alone. No pitch, and nothing to sign - you decide whether you want to take it further.

Book a free 30-minute call

Or just ring us on 07816 970 768, or email hello@seamwork.co.uk.

We reply within a working day. If we cannot help, we will tell you straight away - no pressure, no commitments.

We explain what we do with this in our privacy notice.