Straight answer: Most small business owners are quietly paying between £150 and £400 a month for AI tools they use maybe twice a week, and almost nobody talks about this because "AI saves you money" is a much better story than "I got seduced by eleven free trials and forgot to cancel three of them." I added up my own stack, cut it in half, and I'm going to show you the exact numbers.
The morning I added it all up
Back in October I sat down with my bank statement and my Notion page of "tools I'm trying" and did something I'd been avoiding for months: I added up every AI subscription hitting my card. I run a small consultancy, not a big agency, so this isn't enterprise spend. This is one person, plus a part-time assistant, trying to keep up.
The total came to £223 a month. Not one big invoice, which is why nobody notices it. Seven small charges scattered across the month, each one looking harmless on its own.
- ChatGPT Plus: £20
- Claude Pro: £18
- Descript (AI video and audio editing): £22
- Canva Pro with Magic Studio: £10
- Zapier Starter (for automating AI workflows): £19.99
- Fireflies.ai (meeting transcription and summaries): £10
- An AI writing tool I'd signed up for during a launch offer and completely forgotten about: £39
- A voice cloning tool I used once for a client video: £29
- HeyGen for AI avatar videos, used exactly twice: £24
That last three add up to £92, and I could not have told you their names from memory before I opened my banking app. That's the uncomfortable bit. It's not that AI tools are bad value. It's that most of us are running a subscription business inside our own business, and nobody's watching the till.
Why this happens (and why the "AI saves money" pitch skips it)
Every AI vendor pitch tells you the same story: this will save you ten hours a week, this replaces a £30-an-hour freelancer, this pays for itself instantly. And sometimes that's true. But the pitch never mentions the tenth tool you'll sign up for six months later because someone on LinkedIn showed a slick demo, and it never mentions that you'll forget to cancel the trial that auto-renewed into a full year plan.
I've watched this happen to clients over and over. A hairdresser I worked with last year was paying for three separate AI scheduling assistants at once because she'd tried each one, kept the free trial running "just in case," and never went back to switch them off. That's £87 a month for tools doing the exact same job, none of them cancelled because cancelling felt like admitting she'd wasted money, so instead she quietly wasted more.
This is the bit most articles about AI for small business skip entirely, because it's less exciting than "AI will 10x your output." The truth is that subscription creep is the new agency retainer trap. Ten years ago small businesses got locked into £2,000-a-month marketing retainers for work nobody checked on. Now the same thing happens in £10 and £20 increments, spread across six tools, which somehow feels less painful even though the total is worse.
The 20-minute audit that fixed it for me
I didn't need a spreadsheet consultant for this. I needed twenty minutes and my bank statement. Here's exactly what I did, and what I'd tell any small business owner to do this week:
- Step 1: Pull the last two months of card statements and highlight every line that says "AI" or a tool name you recognise. Don't trust memory, use the statement.
- Step 2: Next to each one, write the actual number of times you used it in the last 30 days. Not "I use it sometimes." A real number.
- Step 3: Anything used fewer than four times in a month gets cancelled that day, not "next month when I have time." You never have time next month either.
- Step 4: For tools that overlap (I had three doing meeting notes), keep the one your whole team opens, not the one you personally like best.
- Step 5: Set a calendar reminder for the 1st of every month to repeat this in five minutes flat. It stops the creep coming back.
After that audit I cancelled the forgotten writing tool, the voice cloning tool, and HeyGen. I kept ChatGPT Plus, Claude Pro, and Fireflies. My new monthly total is £48. I lost nothing I was using. I just stopped paying for hope.
What's worth the money (and what isn't)
Here's where I'll be blunt because most people selling AI tools won't be: not every AI subscription pays for itself, and the ones that do are usually the boring ones, not the flashy ones.
Claude and ChatGPT earn their keep for me because I use them daily for drafting, restructuring client proposals, and thinking through strategy out loud. Fireflies earns its keep because it turns every client call into a written summary I can search later, which has saved me real embarrassment more than once when a client insisted they'd never agreed to something and I could pull up the exact line from the transcript.
The tools that didn't earn their keep were the ones I bought because they looked impressive in a demo, not because I had a repeating problem they solved. AI avatar video is a good example. It's clever technology, but I make maybe one avatar video a year, so paying £24 a month for it is like renting a van every day because you moved house once. If you're a company that consistently creates short, repeatable video content the way Duolingo does, that math changes completely. For most one-person or small-team businesses, it doesn't.
The same logic applies to automation platforms. Zapier is brilliant if you have a workflow you run every single day, connecting your form submissions to your CRM to your email tool. If you set it up once and never touched it again, you're paying £20 a month for a Rube Goldberg machine nobody's watching. The businesses that get this right, the ones like Hotjar built their brand around, are the ones that treat automation as infrastructure they check monthly, not a magic trick they set and forget.
The bit nobody wants to say out loud
Here's the uncomfortable truth: buying more AI tools is often a way of avoiding the harder decision, which is hiring a person or getting proper help to design the system in the first place. I see small business owners add a fourth AI tool to solve a problem that a two-hour conversation with someone who's done this before would fix in one sitting. The tool feels like progress because you did something. The conversation feels like admitting you're stuck.
If you've got three or more overlapping AI subscriptions and you're still not sure your marketing or your automation works, that's usually a sign you need a person to look at the whole picture, not another £15 tool. That's exactly the gap an AI consultant for a small business is meant to close: someone who sits down, looks at what you're trying to achieve, and tells you which two tools do the job instead of letting you collect nine.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
And if the worry is cost, it's worth checking what that kind of help runs before assuming it's out of reach, because the numbers are usually far closer to your monthly tool spend than people expect, and you can see real figures on what an AI consultant typically costs.
A simple way to think about new tools going forward
I now use one rule before signing up for anything new: I have to name the specific, repeating task it solves, not the general category. "It'll help with marketing" doesn't count. "It writes the first draft of my weekly newsletter every Tuesday" counts. If I can't name the exact recurring job, I don't buy it, no matter how good the demo looked.
This is the same discipline that separates businesses that use AI well from businesses that just spend money on AI. Revolut's marketing team didn't win by adopting every new platform going. They picked a small number of channels and tools and went deep on them. Calm did the same thing with content, staying narrow and consistent rather than chasing every format. Small businesses can copy that instinct even without the budget: fewer tools, used, beats a drawer full of half-used trials.
One more thing worth trying if you're deciding whether a paid AI tool is worth the money before you commit: build a rough version of the return yourself. I've had clients use a simple interactive calculator on their own site to work out time saved versus subscription cost, and the same thinking applies in reverse, when you're the one buying the tool rather than selling one.
There's also something Dale Carnegie wrote almost a century ago that I think about every time I'm tempted by a shiny new AI dashboard: he talked about the value of genuine interest over cleverness, of understanding what a person needs rather than showing off what you can do. Half the AI tools on the market are built to impress you in a demo, not to quietly solve your Tuesday problem. The lessons from Dale Carnegie on plain, useful value still apply, even when the tool in question is built on a language model instead of a handshake.
What I'd tell you to do this week
Open your bank statement. Find every AI charge. Write down the last time you used each one. Cancel anything you can't remember using in the past two weeks. Keep a small number of tools you use daily and know well. Set a monthly reminder to do it again. That's it. It took me twenty minutes and cut my monthly AI spend by £175. No new tool required.
Frequently asked questions
How much should a small business realistically spend on AI tools each month?
For a solo business or a team of two to five people, £40 to £80 a month is usually enough to cover one strong chat assistant, one transcription or meeting tool, and one design or writing tool used daily. Anything much beyond that should map to a specific, repeating task, not general enthusiasm for AI.
Why do small businesses end up paying for so many overlapping AI tools?
Free trials convert to paid plans automatically, demos are designed to impress rather than inform, and cancelling feels like admitting a mistake, so people keep tools running out of quiet guilt rather than actual use. It creeps up in £10 and £20 increments rather than one obvious bill, which is exactly why it goes unnoticed for months.
Is it better to buy more AI tools or hire help to set them up ?
If you already have three or more overlapping tools and you're still unsure your setup works, that's usually a sign you need someone to look at the whole system rather than another subscription. A short paid session with someone experienced often replaces months of trial and error, and it tends to cost far less than people assume once they check real figures.
What's the quickest way to check if an AI subscription is worth keeping?
Name the exact recurring task it does, not the general category. If you can say "it drafts my Tuesday newsletter" or "it summarises every client call," keep it. If the best answer you can give is "it helps with marketing" or "it seemed useful," cancel it and revisit the idea only when you have a specific job for it to do.
Related reading: How to Find Out Who Viewed Your Instagram Profile (What Happens When You Try) and How to See Who Unfollowed You on Instagram (Real Methods That Work in 2026).
