GRACE: The travel-guide site did what a lot of sites tried, generated hundreds of articles with AI, published them fast to cover every destination, and traffic is now falling instead of climbing. The assumption was that volume of content equals volume of traffic. What got produced was volume of pages, which is not the same thing, and search engines have gotten good at telling the difference.
HANNAH: Let me be precise about what is and isn’t the issue, because “AI content” gets blamed as if the tool is the problem. It isn’t, the method of production isn’t what search engines judge, the value of the result is. The framework search engines push here is people-first content, made to help a person rather than to game a ranking, and AI-assisted content that’s genuinely useful, accurate, and original clears that bar fine. The problem here is mass-produced pages that say what every other generic travel page says, no firsthand knowledge, no specifics, nothing a hundred other sites don’t already have. The issue isn’t that it’s AI, it’s that it’s generic and redundant at scale.
MARCUS: Hold on, I want to challenge the volume framing itself, because even excellent travel content at that scale might be the wrong play.
HANNAH: Meaning the strategy, not just the quality.
MARCUS: Right. A new or mid-size travel site publishing hundreds of pages overnight is competing head-on with established destination sites that have years of authority and real on-the-ground content. Even good pages, at that volume and that speed, look like exactly what they are, a content dump, and they dilute the site’s overall quality signal. The question isn’t only “is each page good,” it’s “should this site be spreading thin across hundreds of topics instead of being genuinely excellent on a focused few.”
GRACE: That’s the harder truth, the volume strategy might be wrong even if the writing were good, which here it isn’t.
ELENA: And there’s a structural cost to the dump that compounds the quality problem. Hundreds of thin, similar pages don’t just fail individually, they drag the whole site, because search engines assess quality at the site level, not only page by page, the framework here is the helpful-content idea that a large proportion of unhelpful pages can suppress how the whole domain is judged. So the weak pages aren’t neutral, they’re actively pulling down the pages that might otherwise have done fine. The mechanism is site-wide, which is why removing the dead weight can lift the survivors, the dump doesn’t just not help, it harms what was already there.
NOAH: The pattern is treating content as a quantity to manufacture rather than value to create, the same more-equals-better reflex, now turbocharged by a tool that makes producing pages nearly free. When production gets cheap, the temptation is to produce more, and the constraint that used to force selectivity, effort, disappears. The tell is a sudden burst of hundreds of pages, volume that no genuine expertise could have backed in that timeframe.
SOFIA: And the reader feels the emptiness immediately, which is what search engines are reading through behavior. Someone lands on one of these guides looking for real help with a destination and gets the same generic overview they could get anywhere, no specific tips, no current detail, nothing lived. They bounce, they don’t come back, and that pattern across hundreds of pages tells search engines the content doesn’t satisfy. The traffic drop is users voting with their behavior, aggregated.
THEO: So the rule is to optimize for value per page, not pages per month, and to fix the existing damage through a content audit. Pull every dumped page against its actual performance, the impressions, clicks, and engagement it draws, and sort each into keep, improve, or remove, the pages with real signs of life and a distinctive angle get improved, the pages drawing nothing and saying nothing get removed or consolidated, this pruning is what stops them dragging the site. Going forward, produce content that has something the generic version doesn’t, firsthand detail, specific current information, a real angle, whether or not AI assists the drafting. Fewer pages that each earn their place beats hundreds that don’t.
AIKO: And operationally the lesson is that cheap production needs more editorial discipline, not less, because the tool removed the natural brake. The workflow needs a value bar every page must clear before publishing, does this offer something the existing results don’t, plus a periodic audit catching low-value pages before they accumulate into a drag. Using AI to draft is fine, using it to bypass the question of whether a page should exist is what caused this. The discipline moves from “can we produce it” to “should this exist and what makes it worth ranking.”
GRACE: Which is where I’d close, the falling traffic isn’t punishment for using a tool, it’s the predictable result of publishing volume with no value behind it, and the fix is to make each page worth a visit, not to make more of them.
DANA: Here is the decision, value per page over page count, with the existing dump repaired. Traffic is dropping not because the content is AI-assisted but because it’s generic and redundant at scale, and a high ratio of low-value pages drags the whole site, so the weak pages are actively harming the rest. Per Marcus, the volume strategy itself may be wrong, a site spreading thin across hundreds of topics loses to focused excellence, so we audit what’s published, improve the pages that can be made genuinely distinctive, and remove or consolidate the generic remainder that’s pulling the site down. Going forward, every page clears a value bar, does it offer something the existing results don’t, with AI assisting the draft but never bypassing whether the page should exist. The instinct to cover topics was fine. Mistaking a pile of generic pages for coverage is what dropped the traffic.
SOFIA: Search rewards a page that helps someone, not a page that exists. Hundreds that say nothing add up to a site that says nothing.
DANA: The tool made pages cheap, which makes judgment the scarce thing. Publish what genuinely helps, repair what doesn’t, and stop counting pages as if they were value.