The short version: AI tools are trained on the same data, so left alone they produce the same rhythm, the same "here's the thing" openers, the same three-word punchy sentences, on every account using them. I audited fifty LinkedIn posts from small business accounts using AI and found the pattern was obvious within the first ten. The fix isn't a cleverer prompt, it's putting hard limits on what the AI is allowed to touch.
Worth reading next: Why Your AI Content Sounds Like Everyone Else's (And How I Fixed Mine).
The newsletter that made me stop and reread it
Six months into rebuilding my business in public, I asked ChatGPT to draft the intro to my Tuesday newsletter. I gave it my usual brief: warm, blunt, first person, no fluff. It came back with something decent. Too decent. I read it twice because something felt off, and then I realised what it was: I had read that exact rhythm somewhere else that week. Short punch line. Colon. Three examples. Another short punch line to close the paragraph.
I went looking for where I'd seen it. Found it in a newsletter from a marketing consultant in Manchester. Found a version of it again in a LinkedIn post from someone selling AI training in Texas. Same structure, different products. That's when I stopped treating this as a one-off and started treating it as a pattern worth measuring.
The audit: 50 posts, one pattern
I pulled fifty LinkedIn posts from small business owners and consultants who openly say they use AI for content, a mix of AI marketing accounts, coaches, and freelance writers. I wasn't guessing which ones used AI, most of them say so in their bio or their captions. I read them cold, one after another, and logged three things: opening line structure, sentence length pattern, and closing move.
- 38 out of 50 opened with either a rhetorical question or a short standalone sentence under six words
- 31 out of 50 used a "not X, but Y" construction at least once in the first three paragraphs
- 29 out of 50 closed with a call to action phrased as a question ("What's stopping you?" or a close variant)
None of this is a scandal. It's what happens when thousands of people ask the same tool, trained on the same internet, to sound "punchy" and "authentic". The tool has one idea of what punchy sounds like, and it hands that idea to everyone who asks. I do the same thing on a smaller scale when I'm tired and let a draft run without editing it hard. I've caught myself starting posts with a rhetorical question three times in one month before I noticed.
Why the standard advice doesn't fix it
Most posts on this topic tell you to "add your personal stories" or "give the AI more context about your brand". I tried both, for weeks. It helps a little. It doesn't solve the actual problem, because the issue isn't a missing detail, it's the shape of the sentence underneath the detail. You can feed AI your most personal story and it will still wrap that story in the same three-beat structure it wraps everyone else's story in. The content changes. The skeleton doesn't.
Here's the uncomfortable part nobody selling AI courses wants to say out loud: if you let AI write the full first draft of everything you publish, your writing will drift toward the average of everyone else using the same tool, no matter how much personality you try to bolt on afterward. It's not a prompt problem. It's a structural one. The tool is designed to predict the most likely next word based on what's already been written a million times, so the more you rely on it for full drafts, the closer you sit to the middle of the pack, which is the opposite of what you're trying to do in a crowded market.
Brands that stand out don't get there by sounding cleverer, they get there by sounding consistently like one specific person or one specific point of view over years. Look at how Mailchimp built its slightly odd, illustrated, self-aware voice long before AI writing tools existed, or how GoPro built a brand around other people's footage instead of its own copy. Neither of those came from a template. They came from someone deciding, repeatedly, what the brand would never sound like.
The three changes that worked
I didn't stop using AI. I use it every day, for research, for first-pass structure, for pulling ideas out of a messy voice note I've recorded in the car. But I changed how it's allowed into the process, and it made a real difference within about three weeks.
- AI never writes the first or last paragraph. Every opening and closing line I write myself, by hand, before I let AI touch anything. Those two spots carry almost all of a piece's personality, and they're the cheapest to protect.
- I ran my own drafts through an AI detector for a fortnight, as an experiment, not a rulebook. Out of 30 drafts where I'd let AI write more freely, 26 scored above 80% likely AI-generated. After I applied the hand-written opening and closing rule and cut every "not X, but Y" construction on sight, that dropped to 6 out of 30 scoring above 20%. I'm not precious about detector scores as a measure of quality, but the shift told me the structural fix was working, not just a feeling.
- I banned myself from rhetorical questions as openers for a full month. Harder than it sounds. I had to physically stop, delete the question, and write a flat statement instead. It felt clunkier at first. Readers told me it felt more direct.
None of this is about writing "more human" content in some vague sense. It's about removing the three or four specific tics that make content sound machine-average, and doing it on purpose rather than hoping personality leaks through on its own.
Distinctiveness isn't a writing trick, it's a decision made early and repeated
The businesses that keep their voice through the AI era aren't the ones with the best prompts. They're the ones who decided years ago what they will never do, and stuck to it. Figma built its whole marketing tone around a design community that talks to itself, not around clever copy. Instagram's early growth had nothing to do with words at all, it was a format decision, repeated relentlessly. Joe Pulizzi, whose content marketing thinking I've quoted more than once, built his career on one unfashionable idea repeated for a decade rather than chasing whatever format was trending that quarter, a lesson I go back to in what his career teaches about staying power. None of that is replicable by a better prompt. It's a stance, held over years.
If you want a concrete tactic that AI can't flatten into sameness, build something interactive that's specific to your own numbers, not your words. A calculator, a quiz, a pricing tool. I've written before about how interactive calculators convert better than static content precisely because they can't be templated the same way twice, and that's still true now that everyone's writing sounds alike. The output depends on the visitor's own inputs, so it can't average out to the same paragraph as your competitor's.
If you're running a one-person business from home, this hits harder
Solo founders and freelancers are the group I worry about most here, because you're often the whole content team, using the same two or three AI tools everyone else in your niche is using, with nobody in the room to say "this sounds like everyone else's newsletter this week". I know this because I am that person most mornings. There's no second pair of eyes unless you build one in deliberately.
Three practical habits that help when you're working alone:
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.
- Read one competitor's AI-assisted content before you write your own, not to copy it, but to notice the tics it's using so you can deliberately avoid them that day
- Keep a running list of phrases you've caught yourself repeating and ban each one for a month at a time, then rotate
- Record yourself talking through an idea for two minutes before you open any AI tool, and use that recording, not a typed prompt, as the seed for the draft. Speech patterns resist the average far better than typed instructions do
If this is the bit you're stuck on, whether AI is helping your marketing or quietly making it forgettable, it's worth getting a second opinion from someone outside your own head. That's a lot of what I do in client work now, sitting with a small business's existing content and working out which three habits are making everything sound like a stock photo, which is the kind of audit I run through hands-on AI consulting for small businesses rather than in a generic course.
What I'd tell you if you only remember one thing
AI didn't ruin anyone's voice on its own. It just made the average easier to reach and cheaper to produce, which means the average is now everywhere. Standing out in 2026 isn't about finding a better AI tool than the next person, it's about deciding, on purpose, which parts of your writing you will always do yourself, and never handing those parts over no matter how good the draft looks.
Free resource: The Newsletter Growth Mini-Guide.
Free resource: The AI SEO Brief Prompt Pack.
Frequently asked questions
How can I tell if my content sounds too much like AI?
Read your last five posts back to back and check for repeating patterns: rhetorical question openers, "not X, but Y" constructions, and questions as closing lines. If three or more of your last five pieces share the same structure, that's the tell, not any single word choice.
Should small businesses stop using AI for marketing content?
No. Use it for research, outlines, and first drafts of the middle section, but write your opening and closing lines yourself every time. That single rule protects most of the personality a piece needs without slowing down the process much.
Does using AI detectors help improve writing quality?
Not as a quality measure on their own, detectors just flag statistical patterns. But used as a rough before-and-after check while you change your process, they can show whether a structural fix is working, which is how I used them here.
What's the fastest fix if I only have five minutes before publishing?
Delete the first and last sentence of the draft and write both fresh, by hand, in your own words. Those two spots carry most of the tone, and rewriting just those two often does more than editing the whole middle section.
Related reading: The AI Sameness Problem: Why Your Small Business Marketing Sounds Like Everyone Else's and How I Turn One Client Result Into a Month of Content With AI (Without It Reading Like Everyone Else's).
I go much deeper on this in the AI marketing guide.
This goes deeper into practical AI training for teams.
