When ChatGPT launched in late 2022, our phones stopped ringing.
Not all at once. But over the next year, client after client told us the same thing. They were moving to AI-generated content. It was faster, cheaper, and “good enough.” Why pay human writers when a prompt could do the job in seconds?
We lost roughly 80% of our business.
We kept hiring writers anyway.
Our writers are independent contractors. When clients disappeared, so did the work we could send them. But we did not pivot to AI. We kept building a roster of human writers because we watched what happened to the companies that made the switch. Their blogs filled up fast. Hundreds of posts. Thousands of keywords targeted. But the traffic did not follow. Rankings did not improve. In many cases, they got worse.
We did not have the data to prove our bet was right at the time. Now we do.
The Research That Caught Our Attention
Graphite, a research-driven growth agency, published a study that analyzed over 31,000 keywords across Google Search, ChatGPT, and Perplexity. They wanted to know a simple thing. When people search for something, is the content they find written by humans or generated by AI?
The answer was not close.
86% of articles ranking in Google Search are written by humans. AI content accounts for 14%.

Over in the answer engines, the split is similar. ChatGPT and Perplexity cite human-written articles 82% of the time.
And it gets worse for AI content. At the number one position in Google, only 7% of articles are machine-produced. AI content does not just show up less often. It shows up lower on the page.
More Content, Fewer Results
The part that surprised us most is the supply side. A separate Graphite study found that AI articles now outnumber human ones on the open web. Companies are publishing more AI content than ever before.
But Google is not rewarding it. ChatGPT is not citing it. Perplexity is not linking to it.
Think about what that means. The majority of new content being published is AI-generated, and search engines are effectively ignoring most of it. The companies that fired their writers to save money are spending time and resources on content that nobody finds.
What We Think Is Happening
We work with content every day. We read what our clients’ competitors publish. We see the patterns.
AI-generated content tends to be correct but generic. It covers the topic without adding anything new. It reads like a summary of other summaries. Search engines have gotten good at spotting that, whether through AI detection or through engagement signals that tell them users are not finding what they need.
Human writers bring something different. They bring experience with the subject. They have opinions. They write in a voice that does not sound like every other article on the same topic. They can tell you what they tried and what failed, not just what the best practices say.
That is hard to automate. And apparently, search engines can tell the difference.
The Middle Ground Worth Talking About
Graphite noted that they did not study AI-assisted content, where a human writer uses AI to draft or outline and then rewrites substantially. They said this approach may be effective.
We agree. Several of our writers use AI tools in their workflow. They use them for research, for brainstorming angles, for generating rough outlines. Then they throw away most of it and write the piece themselves.
The tool is not the problem. Removing the human is.
What We Tell Clients Now
When someone comes to us after a year of AI-generated content that did not perform, the conversation is usually the same. They published a lot. Rankings went sideways or down. Traffic did not grow the way they expected. They are ready to try something different.
We do not lecture them about it. The pitch for AI content was convincing. The cost savings were real. The quality looked good enough on the surface.
But “good enough” is not good enough for Google. Not when the algorithm is choosing human content 86% of the time.
Our pitch is simple. We have over 40 writers who create content that search engines and answer engines consistently prefer. Not because we have cracked some secret formula. Because we put real people behind every piece, and the data shows that is what works.
If your AI content experiment did not deliver, you are not alone. And we are still here.
Need content like this? Let our writers handle it
Our Blog Writing Services →Frequently Asked Questions
We lost roughly 80% of our business.
The answer was not close.
Not all at once. But over the next year, client after client told us the same thing. They were moving to AI-generated content. It was faster, cheaper, and "good enough." Why pay human writers when a prompt could do the job in seconds?
We do not lecture them about it. The pitch for AI content was convincing. The cost savings were real. The quality looked good enough on the surface.




