Straight answer: Your emails, LinkedIn posts and website copy sound like everyone else's because you and your competitors are typing the same three prompts into the same tools and hitting send on the first draft. The fix isn't avoiding AI, it's feeding it your own words first. Build a small "voice bank" from your real sentences, run everything through it, and you'll sound like you again within a week.
Related reading: How to Create a Community in WhatsApp Business (Step by Step, With the.
The pattern I started seeing in every inbox
Last autumn I was reviewing outreach emails for three separate clients in one week, an accountant in Manchester, a physio clinic in Tel Aviv, and a small marketing agency in the US. None of them knew each other. All three emails opened with some version of "I hope this email finds you well" and closed with "please don't hesitate to reach out." One had the phrase "in today's fast-paced world" in the second line.
That's not a coincidence, that's ChatGPT's default register showing up in three inboxes on three continents. I've now seen the same rhythm, three short punchy opening lines, a rhetorical question, then a bulleted list with an emoji at the start of each bullet, on maybe forty client pieces this year. It's become the visual signature of "written by AI in under two minutes," and your customers can spot it faster than you think.
Here's the uncomfortable bit nobody selling AI courses wants to say out loud: the tools aren't the problem. A hammer doesn't care what you build with it. The problem is that busy business owners take the first draft, change one word, and post it. I've done it myself, twice, before I caught myself and went back to fix it. The output wasn't bad. It just wasn't mine, and readers who've followed me for years noticed within a day.
Why "just add personality" doesn't work
Every AI marketing guide tells you to "add your personality" to the prompt. I've tried this, dozens of times, with clients who write things like "make it sound witty and warm, British, a bit blunt." The output still lands somewhere between a Hallmark card and a corporate LinkedIn post, because the model has no actual sample of your writing to pull rhythm, sentence length or specific phrases from. Telling ChatGPT you're "blunt" produces generic blunt. It doesn't produce your blunt.
What changes the output is feeding it your own sentences as raw material, not instructions. This is the same lesson Dale Carnegie was teaching about people a century before AI existed, specificity beats generic advice every time, because generic advice describes everyone and therefore nobody.
The voice bank method (the five-minute fix)
This is the exact process I now run with every client before we let AI near their marketing copy. It takes about 45 minutes once, and roughly five minutes a week after that.
- Step 1: Pull 20 real sentences. Go through your last twelve months of emails, voice notes transcribed, sales calls, or WhatsApp replies to customers. Copy 20 sentences you wrote, in your actual voice, good or messy. Not marketing copy, real replies to real people.
- Step 2: Paste them into a document titled "voice bank." No editing. Include the weird bits, the sentence fragments, the way you say "look" or "honestly" or "right, so."
- Step 3: Add three "never" rules. Mine are: never open with a question, never use the word "unlock," never use three-item lists as a crutch for every single point. Yours will be different. Write them down.
- Step 4: Paste the whole voice bank into every AI conversation before you ask for a draft. Literally start the prompt with "here are 20 sentences I've written, match this rhythm and word choice" and then give your topic.
- Step 5: Read the output out loud before posting. If you wouldn't say it to someone's face at a networking breakfast, cut it. This single step catches 90% of the sameness.
One client, a bookkeeper with 340 LinkedIn followers, ran this for six weeks. Her post engagement went from an average of 3 comments to 22 comments on her best week, not because the content changed dramatically, but because three separate people told her "this finally sounds like you." That's the whole game. People don't buy from AI. They buy from a person who sounds like a person.
What the big brands do that you can copy for free
The brands that get quoted, screenshotted and shared aren't the ones with the biggest budgets, they're the ones with a voice so consistent you could recognise a sentence out of context. Look at how Duolingo built an entire personality around one unhinged owl, or how Calm stayed deliberately quiet and slow in a marketing world obsessed with noise. Neither of those brands lets a generic AI draft go out the door. They both have documented voice guidelines that are more specific than "friendly and professional," things like exact words that are banned, exact sentence lengths, exact jokes that are allowed.
Hotjar's whole content strategy is built on saying the plain, slightly awkward truth instead of the polished version, which is precisely what AI defaults away from unless you fight it. And Revolut's marketing works because it sounds like one sharp person talking to you, not a committee. That's the standard to aim for with your voice bank, not perfection, one recognisable person.
The uncomfortable truth about "humanizer" tools
There's a growing industry of tools that promise to "humanize" AI text so it slips past detectors and sounds natural. I've tested several of them on client copy. They don't fix the sameness problem, they just rearrange the same generic sentence into a slightly different generic sentence. Your readers were never running your email through a detector anyway, they were reading it with their own ears, and a shuffled version of the same hollow sentence still sounds hollow.
The real tell isn't grammar or word choice, it's the absence of anything specific. Real writing has a client's actual name, a real number, a real date something went wrong, a genuine opinion that could annoy someone. AI's default output avoids all of that because it's trained to be inoffensive and broadly applicable. No humanizer tool will insert the fact that your delivery was three days late in March and you personally called the customer to apologise. Only you have that sentence. That's the whole advantage you're sitting on and ignoring.
Where this shows up beyond email
The sameness problem isn't limited to newsletters. I'm seeing it on:
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.
- Website homepages, where three different local businesses in the same town now have near-identical "we're passionate about delivering excellence" hero sections.
- Sales proposals, where the structure, headings and even the closing line ("I look forward to the opportunity to work together") are identical across competing bids, which is a genuine problem when a client is reading three proposals back to back and yours reads like a copy of the other two.
- Social captions, where the emoji-bullet-list format has become so common that using it now signals "this account doesn't write its own captions" rather than signalling professionalism.
If you want a quick way to add genuine differentiation to a site rather than more generic text, this is exactly where something like an interactive calculator or tool on your site beats another paragraph of copy, it gives visitors something to do rather than something more to read that sounds like the last five sites they visited.
What I do differently now
I still use AI every single day, for first drafts, for restructuring a messy voice note into paragraphs, for finding the gap in an argument I've made. What changed is the order of operations. I write the ugly first version myself, in my own words, badly if needed, then I let AI tidy structure and catch repetition. Not the other way round. When AI writes the first draft and I edit lightly, the piece drifts toward generic. When I write the first draft and AI tidies, the piece stays mine and gets sharper.
This one change, writing first and polishing with AI second instead of prompting first and editing lightly, is the single biggest lever I've found in five years of doing this work in public. It costs nothing. It just takes longer than copy-paste-send, which is exactly why most small businesses skip it and end up sounding like their competitors by accident.
If your team is producing a lot of AI content across email, social and sales and you can't tell anymore whether any of it sounds like your business, that's usually a sign it's worth getting outside eyes on it rather than guessing. This is one of the more common things I get brought in for through AI consulting work with small businesses, an audit of what's currently going out under your name and a plan to get the voice back before customers notice it's gone.
A simple test to run this week
Take your last five outbound emails or posts. Read them without the sender's name attached. If you can't tell whether it was written by you, your competitor down the road, or a stranger in another country using the same AI tool, that's your answer. Rebuild your voice bank, run the process above, and check again in a month. The businesses that keep their actual voice through this AI shift are going to be the ones customers remember. Everyone else is going to sound like the same helpful, polite, forgettable assistant, because that's exactly what they let write for them.
Frequently asked questions
Does AI-written content hurt my rankings or sales?
There's no evidence search engines penalise AI-assisted writing directly, but generic, interchangeable copy tends to get less engagement, fewer shares and weaker conversion because readers can't tell you apart from competitors, which hurts sales more than any algorithm does.
How do I know if my AI content sounds generic?
Read it without your name attached and ask whether a stranger could tell it was written by your business specifically. If it could belong to any competitor in your industry, it's generic, regardless of whether AI wrote it or you did.
Is it worth paying for AI detection or humanizer tools?
Detection tools are unreliable and frequently flag human writing as AI-generated, and humanizer tools just rearrange generic sentences into slightly different generic sentences. Neither adds the specific detail, real names, numbers, opinions, that makes writing sound human.
How long does it take to fix a generic brand voice?
Building a basic voice bank takes about 45 minutes. Most small businesses see a noticeable shift in tone within two to three weeks of consistently feeding AI their own real sentences before asking for drafts, and audiences usually notice within days.
For the bigger picture, see my full guide to AI marketing.
Related reading: How to Get More Views on Instagram Reels (What Moved the Needle for Me) and Why Instagram Reach Dropped and How to Fix It in 2026.
