How I Got Drawn Into Pseudoscience by SEO (and Now Mix It With AI Chatbots)

The story goes like this. I run a SaaS product, the tech is decent, the work is done, and last month when I opened Google Search Console, the traffic curve was flatter than my ECG. I didn’t think much of it at first, telling myself “SEO is just publishing articles, it’ll be fine in six months”—I repeated that nonsense to myself for three years.

Then I noticed something even stranger: my articles started appearing under a box called “AI Overviews” in Google search. Yes, not at the top, but at the bottom. Google now writes a summary itself and places it in front of my link, and users leave after reading that snippet. I spent a week writing a 2,500‑word article, only to see it reduced to three tiny words “source link,” and users have to scroll another screen down to find it.

That made me rethink my whole content strategy. If you’re still pumping yourself up with “content is king,” you should first see who’s actually on the throne now—not your articles, but ChatGPT, Perplexity, and Bing Copilot.

Answer paragraph of roughly 40‑60 words: Modern SEO’s core has shifted from “write an article that ranks #1 on Google” to “make your content citable by AI systems as an answer.” This means you need to focus on structured data, authority signals, and covering user intent, rather than stuffing keywords or chasing word counts. Content that shows up in AI search results tends to be concise, trustworthy, and frequently cited.

The First Truth: You’re Competing With Robots for the Job

I first realized this on a Friday afternoon. I’d just published what looked like a serious article titled “The Complete Guide to SaaS Pricing Strategies in 2026.” It was thorough, covering everything from freemium to perpetual licenses, and I felt great about it.

The next day I tried an AI search tool—I’m not sure whether to call it a search engine or a chatbot these these two are indistinguishable these—by entering “SaaS pricing strategies.” It instantly generated a 300‑word answer, well‑structured, with references. My article wasn’t there. It felt like you spent an entire day hosting a dinner, only for the guests not to show up and you end up ordering takeout.

Later I investigated this phenomenon. Google’s AI Overviews now present search results like ChatGPT: you don’t see ten blue links; you see a paragraph, a summary, an AI‑generated “this is the answer you’re looking for.” Users don’t click; they just walk away.

For content creators, this means your article must play a new role: not a final destination but a data source. Your content needs to be structured, authoritative, and contextual enough for AI systems to easily capture, understand, and cite. It’s not some deep technical feat, but it’s definitely not something you can solve by “writing more keywords.”

The Second Truth: UGC Rises Faster Than Your Blog Posts

If AI search is the first blow, user‑generated content (UGC) is the second punch.

I noticed that some of my competitors weren’t big bloggers at all. They relied on a single forum post or a long user comment, and it ranked above my month‑long, carefully crafted article. I was skeptical, but the data didn’t lie.

Google’s ranking algorithm is now very friendly to UGC. The reason is simple: when a large model sees the same question answered in many different ways thousands of times in its training data, it trusts those answers far more than a solitary brand blog. Users discussing a topic use more authentic, diverse, and context‑rich language.

I ran an experiment: I took a user’s feedback left in my product, polished it a bit, and turned it into a “case story” post on my site. That single post generated more organic traffic than the three formal articles I published that same week.

It’s ironic. You think you’re a professional writer, but a casual user’s off‑the‑cuff remark can be more effective than three meticulously crafted versions.

The Third Truth: AI Tools Changed My Entire Workflow—But Not My Pain

Since the trend is obvious, I couldn’t just ignore it. Early last year I decided to adjust my strategy, stop battling my own perfectionism, and start experimenting with automation tools to fill the inefficient gaps.

I tried having ChatGPT write articles directly. It was useful, but there were big problems—its output looked fine, but something always felt missing. Missing was “my” experience, the mistakes “I” made, the frustrations “I” vented about. So I restructured my workflow into a semi‑automatic process: AI helps me pick topics, outline, and draft; I fill in the details only someone who has actually done the work knows. For example, “On the first day after this feature launched, we received 13 complaint emails”—AI can’t generate that.

Similarly, I’ve been trying to let it handle repetitive but unavoidable tasks: topic research, competitor tracking, scheduled publishing—things I used to do manually, which turned out to be stupidly inefficient. I’m basically a high‑end task scheduler.

At that point I started a small‑scale trial of a tool called SEONIB, which primarily helps me link trend discovery to multi‑platform publishing. Initially I used it just for scheduled posts, but then I discovered it could automatically generate buyer guides and tutorial articles from my product links, saving me a lot of work. I treat it as a helper that can do work automatically, not a “one‑click SEO miracle.” It’s obviously not a cure‑all, but it did pull me out of the daily “what should I write today?” fog. SEONIB handling topic selection and scheduling gave me a genuine sense of relief for two days, then I spent a whole day polishing the details of the article it generated. That’s perfectly normal.

Honestly, for someone like me who loves to nitpick every piece of writing, outsourcing part of the content to a machine always feels like a betrayal. But there’s no choice—machines work faster than me and never stare blankly at an empty screen for half an hour.

The Fifth (Yes, I Skipped the Fourth Because I Forgot While Writing) Truth: Your SEO Stack Needs to Learn New Tricks

The biggest variable in SEO over the past three years wasn’t the search engine; it was “generative engine optimization.” I was stunned the first time I heard that term last year, thinking it was just SEO under a new name.

It’s not that simple.

The underlying logic of SEO hasn’t changed: make the system trust your content. But the “system” now includes not only Google’s crawler but also GPT’s context window, Bing Copilot’s knowledge graph, and Perplexity’s citation chain. The criteria these systems use to evaluate content are no longer about keyword density or backlink count.

A concrete example: I found that when I write the conclusion of an article as a “directly quote‑able paragraph”—a sentence that stands on its own—it appears far more often in AI search results. In contrast, the “three‑paragraph lead‑in” rarely gets cited.

So now I start every article by writing the conclusion. The structure is completely reversed. After finishing, I self‑check: if AI extracts this paragraph as an answer, will it embarrass me? If yes, I rewrite it.

I discovered this trick after six months of experimenting, wasting about twenty‑plus articles, and I’m still fine‑tuning it. It may not work for everyone, but it’s more realistic than any tutorial.

Questions You Might Still Have

Should I Stop Writing Long Articles?

Not necessarily, but you need to redefine the purpose of “long articles.” They’re no longer primarily ranking tools; they’re containers for establishing professional authority. If a 5,000‑word piece gets no citations, it’s better to write a 500‑word piece that gets cited three times. My approach is to pair a long article with a quote‑able summary paragraph, formatted so AI can extract it directly.

What’s the Main Difference Between Generative Engine Optimization and Traditional SEO?

Classic SEO is like setting a trap—waiting for traffic on the search engine’s inevitable path. GEO is more like running a specialty grocery store, explicitly telling AI “take my information and use it.” The technical differences aren’t huge, but the strategic mindset is completely different. The real change is expanding your definition of “user” to now include large language models.

Can I Publish User Comments Directly as Articles?

You can, but don’t be lazy. If users provide valuable content, package it into a more formal, indexable format—no problem. But you must add your own perspective, otherwise you become a “content porter.” My usual method is: a customer says something, I use it as the opening, then expand it with the pitfalls I’ve encountered. This way machines understand it, and people find it valuable.

Can Automation Tools Really Solve SEO?

They can solve part of the problem—topic selection, publishing cadence, standardized formatting—but don’t expect them to produce an article that makes readers think “this author definitely works in the field.” Tools excel at preventing mistakes, not at making you shine. The shining part still has to come from you.

At least, that’s true so far.