The short version: AI writing tools pull from the same training data and the same handful of prompt patterns, so thousands of small businesses are publishing copy that reads like it came from the same person. The fix isn’t ditching AI, it’s feeding it your voice before you ask it to write anything, and refusing to publish a draft that could belong to your competitor.
The workshop that proved it
Last spring I ran a half-day session in Manchester for 40 small business owners, mostly consultants, tradespeople, and independent shop owners. Near the end I gave everyone the same seven-word prompt: “write a tagline for my small business.” I told them to run it through ChatGPT and read theirs out loud.
Within four rounds, we had heard the phrase “your trusted partner” six times, “game-changing” four times, and “elevate your” three times. A plumber from Stockport and a graphic designer from Salford, two businesses with nothing in common, ended up with taglines that started identically. Nobody had copied anyone. The tool had simply done what it does, which is predict the most statistically likely next words based on millions of other business taglines it had already seen.
That’s the bit nobody tells you when they hand you a list of prompts. The prompt isn’t the problem. The absence of anything specific to feed it is.
Why this keeps happening (and it’s not really the AI’s fault)
Here’s the uncomfortable part. Most of this sameness isn’t AI failing you, it’s you stopping at the first draft. I’ve watched clients paste in a rough brief, get something back in eight seconds, and hit publish because it sounds professional and they’re busy. Professional and distinctive are not the same thing, and AI defaults to professional because professional is the safest average of everything it was trained on.
There’s a second, less flattering layer to this. A fair few of the “AI content agencies” that popped up over the last two years are running the same three or four prompt templates across dozens of clients in the same industry. I’ve seen it directly: two accountants in neighbouring towns, both paying a content agency around £400 a month, publishing blog posts with near-identical structure, the same three subheadings, the same closing call to action word for word. Their audiences overlap by maybe 20 miles. Neither of them knows the other exists, but Google, and any customer reading both sites in the same week, absolutely will notice.
Brands that win long term do the opposite of average. Look at how GoPro built its brand entirely around customer-shot footage nobody could fake with a template, or how Figma’s marketing leans into a specific, slightly nerdy design-community tone that no generic prompt would ever produce. Distinctiveness is the whole strategy, not a nice extra.
What sameness is costing you
This isn’t a purity argument about authenticity for its own sake. It’s a business problem with a measurable cost.
- Recall drops. If your homepage headline reads like the next four search results, a prospect comparing options has no reason to remember you specifically.
- Ad costs rise. Meta and Google reward creative that stands out in the feed with lower cost per click. Bland, average copy blends into the scroll and you pay more to be seen.
- Trust erodes slowly. Customers are getting sharper at spotting AI-flat writing, and a growing number say it makes them trust a business less, not more, according to a 2025 Pew Research Center survey on public attitudes toward AI-generated content.
- You train your own team to stop thinking. Once “ask the AI” becomes the default first step, the muscle for writing something only your business could say starts to atrophy.
The five-step fix I use with clients
None of this requires giving up AI tools. It requires putting your own material in front of them before you ask for anything.
Step 1: run the twin test
Type your business into ChatGPT with a plain, generic prompt, no context, and pull up the actual copy from two or three competitors at the same time. Count matching phrases and sentence structures. If you find more than two or three overlaps, your current content is closer to average than you think.
Step 2: build a voice file before you write anything
Paste ten examples of writing that sounds like you, an email you sent a client, a LinkedIn post, notes from a sales call, into a document and feed that to the AI tool first, every single time, before you ask for a draft. Tell it explicitly to match that tone and vocabulary, not to improve it. This single habit does more to fix sameness than any clever prompt.
Step 3: keep a banned word list
Write down every phrase that keeps showing up in your drafts that also shows up everywhere else. Mine currently includes “in today’s fast-paced world,” “elevate your,” “game-changing,” and anything that opens with “imagine a world where.” Ban them from anything published under your name. It’s a two-minute edit that removes 80 percent of the flatness.
Step 4: add one detail AI cannot invent
A real number, a client’s actual result, a date, a place, a specific object. AI can write around a detail once you give it one, but it cannot manufacture your Tuesday client call or the exact figure your last campaign hit. If you want a concrete way to make this unavoidable, build something interactive that forces real inputs, like the interactive calculators approach I’ve written about before. No template can fake a tool that does real maths on a real customer’s numbers.
Step 5: test it blind before you publish
Show a draft to three customers or team members with no branding attached and ask one question: “does this sound like anyone else you know?” If the honest answer is “sort of,” it goes back for another pass. This is the step almost everyone skips because it takes an extra day, and it’s the one that catches the sameness before your audience does.
Where this connects to content strategy more broadly
The businesses that never fall into the sameness trap tend to be the ones already thinking about content as a long game rather than a task to clear off a list. Content marketing pioneer Joe Pulizzi has spent two decades making the same argument in different forms, that a business willing to commit to a specific angle, held consistently over years, beats a business chasing whatever performs this week. I wrote about several of his sharpest points in a piece on business lessons from Joe Pulizzi, and the through-line matters here too. AI can help you produce more, but it cannot decide what your specific angle is. That’s still your job.
The same logic shows up in how brands use visual platforms. Look at how deliberate Instagram marketing strategy gets when it works, or how Mailchimp built recognisable brand quirks into something as unglamorous as email software. None of that came from a default prompt. It came from a decision about what the brand would sound and look like, made before any tool got involved.
If you don’t have time to fix this yourself
I get why most small business owners aren’t going to build a voice file and run blind tests between school pickup and closing invoices. If auditing your own content feels like one more thing on a list that’s already too long, that’s exactly the kind of gap an AI consultant for small business exists to close, someone who sits with your actual writing for a day, builds the reference material once, and hands you a system that keeps working after they’ve gone.
The tools aren’t going anywhere and they shouldn’t. AI has made it possible for a one-person business to publish at a pace that used to need a team. The businesses that will win the next few years aren’t the ones who use AI the most. They’re the ones who make sure what comes out the other end still sounds like nobody but them.
Free resource: The AI Meta Ads Prompt Pack.
Frequently asked questions
Why does AI-written content sound so similar across different businesses?
AI language models predict the most statistically likely next words based on huge amounts of existing text, so without specific input about your business, they default to the same safe, common phrasing everyone else gets. The fix is feeding the tool your own past writing and real details before asking for a draft, not avoiding AI altogether.
Can Google or customers tell when content is AI-generated?
Google doesn’t penalise content simply for being AI-assisted, it penalises low-quality, unhelpful content regardless of how it was written. Customers, though, are getting noticeably better at spotting flat, generic phrasing, and surveys including 2025 Pew Research Center polling show a real chunk of the public says it lowers their trust when writing feels AI-generic.
What’s the fastest way to make AI content sound more like my business?
Paste ten examples of your own past writing, emails, social posts, sales messages, into the AI tool and tell it to match that tone before you ask for anything new. This single step fixes more sameness than any prompt trick, and it takes about ten minutes to set up once.
Should a small business stop using AI for marketing content because of this?
No. The problem isn’t the tool, it’s using it with no input and no editing pass. Businesses that keep AI for speed and drafting but add their own specific numbers, stories, and voice before publishing get the benefit without the sameness problem.
For the practical version of this, see AI consultants for UK small business.
This overlaps with my main post on the topic: ai marketing sameness small business.
Related reading: Why AI Pricing Advice Is Quietly Training Small Business Owners to Undercharge and Why Your AI Meeting Notetaker Might Be Breaking the Law (And Killing Your Sales Calls).
