To be honest, the following words are not easy for me to write. Why? Because I’ve been telling everyone in the past few months and weeks: DON’T use AI-generated content on your website.

I haven’t believed in the hype. I was pretty sure that AI-generated content doesn’t rank on Google, nor would it be cited and referenced by AI search engines themselves. I mean, AI is fed by AI to spill out AI-generated content. That feels like the worst case situation to me on a fundamental level.

But one of our goals here at OtterlyAI is to share our GEO and AI search experiments and learnings. So let me tell you about a recent experiment I did and why it changed how I think about AI content generation.

I expected our human-written landing page to outperform the AI-generated one. We’re a team of marketers and SEO professionals. We know our audience. We craft every sentence with intention.

Turns out, intention doesn’t matter as much as structure.

The Experiment: AI vs Human-Created Landing Pages

We created two landing pages targeting the same audience: agencies looking for AI search monitoring tools.

  • Landing Page 1: Human-Created URL: otterly.ai/agencies Written by our marketing team, using the same process we’ve always used.

We published both pages in January 2026 and tracked performance for 30 days.

Landing Page 1: Human-Created Page: otterly.ai/agencies

Landing Page 2: AI-Created Page: otterly.ai/agencies-ai-search-monitoring

The behind-the-scenes of this experiment

At OtterlyAI, we started creating various persona – and solution-oriented landing pages on our website.

We did this because we fundamentally believe that persona-tailored content will perform well. This type of content will help us be more discoverable when people search for “best tool for agencies” in our category. We’re also pretty sure it will help us perform better on AI search, meaning agencies looking for a great AI search monitoring tool will find us because of that content.

So the experiment was clear:

  1. Publish persona- and solution-oriented landing pages
  2. Those landing pages will rank well on Google and get cited on AI search (e.g. ChatGPT)
  3. Ultimately this will lead to increased visibility on Google + AI search

But here’s the twist. We recently introduced a new free GEO tool called the Landing Page Generator. I wanted to put the system to the test. Which landing page performs better: the one generated by us as a marketing team, or the one created by AI?

How We Measured Success

Before we look at the results, let’s talk about measurement. Here are the KPIs I benchmarked both pages against:

  • Website Citations on ChatGPT: Which page gets cited more often on ChatGPT for our relevant queries?
  • Impressions on Google Organic: Which page gets more organic impressions on Google?
  • Agentic Traffic: Which page gets more engagement from AI crawlers?

Here are the analytics tools we used:

  • OtterlyAI: To track website citations on ChatGPT + AI bot traffic
  • Google Search Console: To measure organic Google performance
  • OtterlyAI Agentic Analytics (currently in closed beta): to measure bot traffic on our website

The Results

The AI-generated page didn’t just win. It dominated.

Google Performance: The AI page earned 34,000+ impressions. The human page earned 10,000. That’s a 3x difference on traditional search alone.

ChatGPT Citations: This is where things got interesting. The AI-generated page received 206 citations in ChatGPT over 30 days. The human-written page? Just 5 citations. That’s a 40x difference in AI search visibility.

Interestingly, the ChatGPT-User Agent shows a less dramatic difference. It crawled the human-written page 387 times versus 901 visits to the AI-generated page.

To me, this suggests: the ChatGPT-User Agent is engaging with both pages, favoring the AI-created one. However, when it comes to citations, the AI-generated page clearly stands out, as the ChatGPT-User Agent deems it far more worthy of citation.

Organic Google ImpressionsWebsite Citations on ChatGPTChatGPT-User Agent Page Visit
Landing Page 1: Human-Created URL: otterly.ai/agencies 10,1025387
Landing Page 2: AI-Created URL: otterly.ai/agencies-ai-search-monitoring34,102206901

Here’s a look at our Google Search Console:

And our Citation Report inside OtterlyAI:

Why the AI Page Won

When I dug into what made the difference, the answer was almost frustratingly simple: more content, more structure, better elaboration.

The AI-generated page wasn’t smarter. It was more thorough. It covered the topic from multiple angles, anticipated follow-up questions, and organized information in a way that AI systems could easily parse and reference.

Our human-written page was tighter, punchier, more “on brand.” It assumed the reader already understood the context. AI search engines don’t make those assumptions.

What This Means for GEO

The shift from traditional SEO to Generative Engine Optimization requires a different mindset. In traditional search, you optimize for rankings and clicks. In AI search, you optimize for citations.

Citations favor depth over brevity. They reward clear structure over clever copywriting. The content that gets cited isn’t necessarily the most elegant; it’s the most useful for an AI to reference when answering a question.

Content Depth

The AI-generated page contained more comprehensive coverage. It addressed the subject from multiple angles and included information that anticipated follow-up questions.

Human copywriting often prioritizes brevity. AI search engines favor completeness.

Structural Organization

The AI-generated page used clearer structural organization with distinct sections, subheadings, and logical information hierarchy. This structure makes it easier for AI systems to parse content and extract relevant information for citations.

Topic Elaboration

Rather than making a claim and moving on, the AI-generated page explained concepts, provided context, and connected ideas. This matches how AI systems build responses. They look for content that fully explains a concept rather than content that merely mentions it.

How to Create More Citable Content

Based on this experiment and our broader analysis of over 1 million AI citations, here’s what works:

Structure content clearly. Use descriptive headings that match how people phrase questions. Organize information in a logical hierarchy.

Provide comprehensive coverage. Cover topics thoroughly rather than briefly. Address related questions and concepts.

Include specific data. Statistics, percentages, and specific figures are more likely to be cited than general statements.

Answer questions directly. Start sections with clear statements of the main point. Provide direct answers before elaborating.

Add FAQ sections. Question-and-answer formats match how AI systems process queries.

The Tool Behind the Test

The AI page was created using our GEO Landing Page Creator. We built this tool after analyzing citation patterns across industries and accounts, encoding what actually works into a system that can replicate it.

You can try it yourself at geo.otterly.ai/geo/ai-landingpage-creator/

The tool generated a page that earned its first ChatGPT citation within 24 hours of publishing. Our carefully crafted human page is still struggling to break into double digits a month later.

The Uncomfortable Takeaway

I spent years honing my copywriting skills. Learning to cut the fluff, sharpen the message, make every word count. Those skills still matter for conversion copy, email campaigns, and brand storytelling.

But for AI search visibility? The data suggests a different approach. More content. Clearer structure. Comprehensive coverage of the topic.

The AI didn’t write better. It wrote differently. And for this particular game, different is winning.


This GEO experiment is part of our ongoing research into AI search optimization. At OtterlyAI, we track hundreds of millions of website citations monthly across ChatGPT, Perplexity, and Google AI Overviews. For more insights, visit otterly.ai.