I'll admit it. I let ChatGPT write my homepage first draft. It took maybe forty seconds. The copy sounded confident, hit every section I asked for, even threw in a headline that made me nod along for a second. Then I read it again the next morning with actual coffee in my system, and I started fixing things. A lot of things.
This isn't a story about AI being useless. It's the opposite, honestly. AI copywriting is a genuinely useful starting point. But starting point is the key phrase here, and if you skip the fixing part, you end up publishing something that reads fine on the surface and falls apart the moment a real customer, or a search engine, actually pays attention.
The First Problem: It Sounded Like Everyone Else
The draft was grammatically perfect. Every sentence made sense. And somehow it still sounded like it could belong to any SaaS company, any agency, any brand with a similar offer. Nothing in it screamed "this is us" instead of "this is a company."
This is the core issue in the ai vs human copywriting debate that people argue about online constantly. AI models are trained on enormous amounts of existing content, which means their default output tends to converge toward the statistical middle of how things get phrased. Safe. Competent. Forgettable. A human writer who actually knows the brand brings in specific details, real anecdotes, phrasing quirks that a model has no way of inventing because it doesn't have lived experience to pull from.
I ended up rewriting almost every headline because they were technically fine and completely interchangeable with a competitor's homepage.
The Second Problem: Vague Claims Standing In For Real Proof
The draft was full of lines like "trusted by businesses worldwide" and "proven results that speak for themselves." Nice rhythm. Zero substance. Nobody had told the model what our actual numbers were, so it filled the gap with confident-sounding nothing.
This is one of the quieter ai copywriting risks that doesn't get talked about enough. AI will generate persuasive-sounding language even when it has no real information to persuade with. It doesn't know that's a problem. It just completes the pattern. A human editor has to go back through and either supply real data or cut the claim entirely, because vague confidence reads as marketing fluff the second a skeptical visitor notices there's nothing behind it.
The Third Problem: Structure That Looked Right But Felt Wrong
Every section followed the same rhythm. Short punchy line, supporting sentence, call to action. Repeated across the whole page like a template stamped out five times in a row. Individually each section was fine. Together they created this strange, flat monotony that made the whole homepage feel mechanical to scroll through.
Real persuasive writing varies its rhythm on purpose. Some sections need to build tension slowly. Others need one sharp sentence and nothing else. I had to go through and manually break that pattern, cutting some sections down to almost nothing and letting others breathe a little longer where the offer actually needed more explanation.
Why This Actually Matters for Search Rankings Too
Here's the part a lot of people miss when they think about ai copywriting purely as a writing quality issue. It's also an SEO issue, and a fairly serious one.
Google has been explicit that it doesn't penalize content simply for being AI-assisted. What it does penalize is content produced primarily to manipulate rankings rather than to genuinely help the reader, regardless of whether a human or a machine typed it. The company's own Search Central guidance on AI-generated content makes this distinction directly, and the broader quality rater guidelines lean heavily on concepts like experience, expertise, authoritativeness, and trust, things that thin, generic AI output tends to lack by default.
So an unedited AI draft doesn't just risk sounding forgettable to a human visitor. It risks looking like exactly the kind of low-value, mass-produced content Google has been actively working to push down in results. The fix isn't avoiding AI. It's making sure whatever gets published clears that quality bar before it goes live.
What I Actually Kept From the AI Draft
I want to be fair here, because I didn't throw the whole thing out. The overall structure of the page, the order of sections, the basic logic of what should come before what, that part was genuinely useful and saved me real time. The first pass at each headline also gave me something to react against, which is often easier than staring at a blank page.
Where AI earns its place in this workflow is speed on the scaffolding. Where it falls short is everything that requires actually knowing the business, the customer, and the details that make a page trustworthy instead of just readable.
The Editing Checklist I Ended Up Using
After going through this once, I built a rough mental checklist for any AI draft before it goes anywhere near a live page. Replace every vague claim with a specific number or cut it. Read every headline out loud and ask if a competitor could publish the exact same line. Break up any section that follows an identical rhythm to the one before it. Add at least one detail a model could not have guessed, something only someone inside the business would know.
That last one turned out to be the single biggest lever for making the page feel real instead of generated.
Bringing It Back to Strategy
If you're rethinking how AI fits into your own content process, it helps to zoom out past just one homepage and look at the bigger picture of traffic and conversion strategy. Tools and resources over at Va Traffic Pro are worth checking out if you're trying to build a repeatable system rather than fixing pages one at a time forever.
The Bottom Line
ChatGPT gave me a homepage in forty seconds. It took me most of an afternoon to make it actually sound like a company worth trusting. That gap, between fast draft and genuinely good copy, is exactly where the real value of a copywriter still lives. AI didn't replace that work. It just moved where the work happens.
FAQs
Does Google penalize AI-generated copywriting?
Not for existing, no. Google's actual position is that it doesn't care whether a human or a model typed the words. What it penalizes is content built mainly to game rankings instead of help someone. A lot of raw AI output ends up in that bucket by accident, not because it's AI, but because nobody bothered to make it actually useful.
What's the real difference between AI and human copywriting?
AI plays it safe. It's trained on millions of examples, so it naturally drifts toward whatever sounds "normal," which is a nice way of saying forgettable. A human writer knows things a model can't. Real numbers. Actual customer objections. The weird detail that makes a brand sound like itself instead of a template.
What should worry you most about AI copywriting?
Confident nonsense, honestly. Lines like "trusted by businesses everywhere" that sound persuasive but mean nothing. Repetitive structure that makes a whole page feel stamped out. And thin content that Google's quality systems are specifically built to catch.
Can AI-written pages still rank?
They can. I've seen it happen. But only once someone's gone through and added real proof, cut the filler, and broken up that robotic same-length-every-time rhythm AI defaults to.
Is editing an AI draft really necessary, or is that overkill?
It's not optional, at least not if you want the page to convert anyone. The draft gets you a skeleton. Everything that makes visitors trust it, specifics, tone, proof, still has to come from a person who actually knows the business.