Bottom line: AI hasn't made your marketing worse, it's made it identical to everyone else's, and that's more dangerous because nobody notices until their open rates and engagement quietly drop. The fix isn't ditching AI, it's a five-step edit that puts your actual voice back on top of it.
The test I ran with three clients
Last autumn I gave three different small business owners the exact same brief. Write a LinkedIn post announcing a new service. Same prompt, word for word, into ChatGPT. A bookkeeper in Leeds, a physio in Manchester, a marketing freelancer in Tel Aviv.
All three came back within minutes with posts that opened with some version of "In today's fast-paced business world" and closed with "What are your thoughts? Drop a comment below." Two of them used the phrase "game-changer." One used it twice.
None of these three people write or talk like that. I know because I've sat in calls with all of them. But the AI output was so smooth and so confident-sounding that all three were ready to post it as is. That's the bit that worries me. Bad writing gets caught. Good-sounding writing that says nothing gets published.
Why this keeps happening
Large language models are trained to predict the most statistically likely next word based on enormous amounts of existing text. That means the default output is, almost by definition, the average of everything already written on that topic. Ask it for a LinkedIn post about a new service launch and it will hand you the most common LinkedIn post about a new service launch, because that's what "most likely" means in practice.
Add to that the fact that most people type the shortest possible prompt (write me a post about X) and you get the plainest possible output. It's not a flaw in the tool. It's exactly what the tool is built to do. The flaw is in stopping at the first draft.
I looked back through six months of posts from small business accounts I follow for research and counted how many opened with "In today's" or "In an era of" or similar. Out of 140 posts, 31 used one of those openers. That's more than one in five, all sounding like they came from the same person, because in a sense they did.
The uncomfortable bit nobody wants to say out loud
Here's the part most people writing about AI and marketing skip over. A lot of small business owners don't want their content to sound different. They want it done. Sounding distinctive takes time, requires you to know your own opinions, and means occasionally saying something that might put a customer off. Sounding like everyone else is safe, fast, and free of that discomfort.
I've watched clients read a generic AI draft, wince slightly, and post it anyway because rewriting it felt like work they didn't have time for. That's not an AI problem. That's a business owner choosing speed over identity, and it's a choice worth naming plainly because it's the real reason the sameness spreads. The tool didn't do this to marketing. Convenience did.
What the sameness costs you
I tracked engagement across one client's LinkedIn page for four months. In the two months where she was using unedited AI drafts, average engagement per post was 38 reactions. In the two months where she used AI for a first draft then rewrote every opening line and closing line herself, average engagement was 67 reactions. Same posting frequency, same topics, same audience. The only variable was whether the voice sounded like her or like the model.
That's not a scientific study, it's one account over sixteen weeks, but it matches what I see across the businesses I work with. People scroll past content that sounds like it could have come from any account. They stop for content that sounds like it came from one specific person with one specific view.
Brands that get remembered do the opposite of average. Look at how Duolingo built its marketing strategy around a slightly unhinged, very specific personality rather than safe corporate copy, or how Calm built its brand on one consistent, recognisable tone across every single touchpoint. Neither of those brands got there by publishing the statistical average of what everyone else in their category was saying.
The five-step edit I use now
I still use AI for every first draft. I'm not precious about that and I don't think you should be either. But nothing goes out the door without this edit, which takes about eight to twelve minutes per post once you've done it a few times.
- Kill the opener. Delete the first sentence entirely. If it starts with "In today's," "In an era of," or "As a business owner," it's the tell-tale sign of an unedited draft. Start instead with a fact, a number, or a sentence someone would only say if they'd done the thing.
- Find the one opinion. AI drafts describe. They rarely take a side. Read your draft and ask: where's the bit I believe that someone else might disagree with? If there isn't one, add it. That's the sentence people screenshot.
- Cut every "game-changer," "unlock," "," and "in the ever-evolving landscape of." These words are load-bearing filler in AI writing. Once you remove them the sentence usually needs rebuilding, which is exactly the point.
- Replace the generic example with your real one. If the draft says "for example, a small business owner might struggle with," swap it for the actual client, the actual number, the actual thing that happened last Tuesday. Specificity is the fastest way to sound like a human rather than a summary of humans.
- Read it out loud. If you wouldn't say a sentence to a client sat across the table from you, cut it. This single step catches more sameness than the other four combined.
Do this consistently and something odd happens after a few weeks. You start writing your own first drafts faster too, because you've trained yourself back into your own patterns rather than the model's.
Where sameness hides beyond social posts
It's not just LinkedIn captions. I've reviewed sales emails from four different consultants this year that all used the phrase "I wanted to reach out" in the first line, all used a three-bullet list of pain points in the middle, and all closed with "Let me know if this resonates." Four different businesses, same skeleton, because all four used the same AI tool with the same default prompt for cold outreach.
Website copy has the same issue. I can usually tell within thirty seconds of landing on a small business homepage whether the "About" section was pasted straight from a chatbot, because it lists generic values (integrity, innovation, customer-first) rather than describing a specific way that business works. Compare that to a brand like Hotjar, whose entire marketing strategy leans on very specific, very plain language about what users see and feel, not abstract values nobody can picture.
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.
Even tone of voice in customer service replies is starting to flatten. I've had support emails from three unrelated companies this year that opened with "I completely understand how frustrating this must be" word for word. That's not empathy, that's the same template running through the same tool.
A quicker way to stand out than fixing every sentence
If rewriting every piece of content feels like too much, there's a shortcut. Build one thing your competitors can't copy from a prompt: an interactive tool, calculator, or quiz that's specific to your business and your numbers. I've written before about how interactive calculators convert more visitors than static content, and part of why they work is that they can't be generated by a chatbot in ten seconds. They require your actual pricing, your actual process, your actual data. That makes them one of the few marketing assets left that's naturally resistant to the sameness problem.
Old thinking still applies here
None of this is new. Dale Carnegie was writing about the same principle nearly a century ago, long before AI existed. People respond to people, not to polished generalities. Some of the business lessons from Dale Carnegie that still hold up today are exactly this: be specific about the other person's situation, use their actual name and actual problem, don't talk in abstractions. AI has just made it far easier to skip that step, because the abstraction now arrives pre-written and sounds confident enough that you're tempted to leave it alone.
The same applies to bigger brands doing this well at scale. Revolut's marketing strategy works because the copy is short, specific, and slightly blunt rather than smoothed into corporate safety. That bluntness is a choice made after the first draft, not the first draft itself.
When it's worth getting outside help
If you've tried the five-step edit for a month and your content still reads like everyone else's, the problem usually isn't the editing steps, it's that nobody's ever helped you work out what your actual voice and opinions are in the first place. That's a slower fix than a prompt tweak, and it's often where working with an AI implementation coach pays for itself, not by writing your content for you but by building the system and the voice guide that stops every draft defaulting back to average.
Frequently asked questions
Is it AI's fault that small business content sounds the same?
Not entirely. The model produces the statistical average of existing writing on a topic by design, but the sameness sticks around because business owners post the first draft without editing it. The tool creates the raw material, the choice to skip editing creates the sameness.
How much editing does an AI draft need?
In my own testing, eight to twelve minutes of focused editing per post, cutting the generic opener, adding one real opinion, and swapping in a specific example, roughly doubled engagement compared with posting the unedited draft, based on a sixteen-week comparison on one client's LinkedIn account.
What's the fastest way to spot AI sameness in my own content?
Read it out loud. If a sentence sounds like something you'd never say to a client sat across the table, or if it opens with "In today's fast-paced world," it hasn't been edited yet.
Should small businesses stop using AI for content altogether?
No. Use it for the first draft, the structure, and the speed. The problem isn't using AI, it's publishing the first output without adding the specific facts, opinions, and examples that only you have access to.
Related reading: Red Bull Marketing Strategy: How They Built a Brand That Wins and Zara Marketing Strategy: How They Built a Brand That Wins.
For the bigger picture, see my full guide to AI marketing.
