AI & Automation

How I Discovered That LLMs Don't Read Websites Like Humans (And Restructured Everything)


Personas

SaaS & Startup

Time to ROI

Medium-term (3-6 months)

I was working on a complete SEO overhaul for an e-commerce client when something weird happened. Despite being in a traditional niche where AI usage isn't common, our content started appearing in ChatGPT and Claude responses. Not something we optimized for - it just happened naturally.

This discovery led me down the rabbit hole of GEO (Generative Engine Optimization), and what I learned completely changed how I think about site architecture. While everyone's still obsessing over traditional search rankings, I realized we're missing a massive shift happening right under our noses.

LLMs don't consume content like search engines do. They break everything into chunks, synthesize answers from multiple sources, and care more about context than keywords. Most businesses are building sites for 2010 SEO when they should be preparing for the AI-first search future.

Here's what you'll learn from my real-world experience:

  • Why traditional page structure fails with LLMs

  • The chunk-level architecture that actually works

  • How to make your content "citation-worthy" for AI responses

  • Practical steps I took to restructure content for both humans and LLMs

  • Why focusing on SEO fundamentals is still your foundation

Industry Reality

What the SEO world is telling you about AI search

Most SEO experts are telling you to ignore the AI search trend. "Focus on users, not algorithms," they say. "Google's still the king." And honestly, they're not wrong about the fundamentals - but they're missing the bigger picture.

The traditional advice looks like this:

  1. Keep doing keyword research - Target specific terms and optimize title tags

  2. Build topic clusters - Create pillar pages and supporting content

  3. Optimize for featured snippets - Structure content for position zero

  4. Focus on E-A-T - Expertise, Authority, Trustworthiness

  5. Don't chase AI trends - Stick to proven SEO fundamentals

This conventional wisdom exists because traditional SEO has worked for decades. Google's algorithm rewards well-structured, authoritative content that serves user intent. The framework is solid and battle-tested.

But here's where it falls short: LLMs consume content completely differently than search crawlers. While Google reads pages linearly and values hierarchical structure, AI models break content into contextual chunks and synthesize information across sources. Your beautiful pillar page might be perfect for Google but terrible for AI citation.

The shift isn't about abandoning SEO - it's about evolving your architecture to work for both traditional search and the growing AI search landscape. Most businesses are going to get caught flat-footed when AI-powered search becomes mainstream.

Who am I

Consider me as your business complice.

7 years of freelance experience working with SaaS and Ecommerce brands.

When I started the SEO project for this e-commerce client, the brief was straightforward: complete site overhaul, improve organic traffic, standard stuff. The client had a solid product catalog but virtually no organic visibility. We were starting from scratch with traditional SEO fundamentals.

The client operated in a pretty traditional niche - not exactly where you'd expect heavy LLM usage. Think practical, everyday products rather than cutting-edge tech. So when we started tracking mentions and discovered our content appearing in AI-generated responses, it caught me completely off guard.

Even though this wasn't a high-tech industry, we were getting a couple dozen LLM mentions per month. People were asking ChatGPT and Claude questions related to our niche, and our content was being cited in responses. This wasn't something we optimized for - it happened naturally as a byproduct of solid content fundamentals.

That discovery made me realize something important: AI search isn't just coming to tech and SaaS companies. It's already happening across every industry, even traditional ones. People are using LLMs for research, comparison shopping, and decision-making in ways that bypass traditional search entirely.

My first instinct was to dive deeper into what was working. Through conversations with teams at AI-first startups like Profound and Athena, I realized everyone is still figuring this out. There's no definitive playbook yet. But what we do know is that the foundation hasn't changed - quality, relevant content remains the cornerstone.

The difference is in how that content gets consumed and cited by AI systems. I needed to understand how LLMs actually process and reference information, then restructure our approach accordingly.

My experiments

Here's my playbook

What I ended up doing and the results.

Instead of abandoning traditional SEO for shiny new GEO tactics, I took a layered approach. The foundation had to remain solid - create genuinely useful content for humans first. But I needed to add a new layer that made that content digestible for AI systems.

The key insight was understanding that LLMs don't consume pages like traditional search engines. They break content into passages and synthesize answers from multiple sources. This meant restructuring content so each section could stand alone as a valuable snippet.

Here's the systematic approach I developed:

Step 1: Chunk-Level Architecture
Instead of thinking in terms of "pages," I started thinking in terms of "chunks." Each section needed to be self-contained with clear context. I restructured existing content so that any paragraph could be extracted and still make sense without surrounding context.

Step 2: Answer Synthesis Readiness
I reorganized information with logical structure that AI could easily extract from. This meant leading with clear answers, then providing supporting details. Instead of burying the key insight in paragraph three, I put it upfront where LLMs could grab it.

Step 3: Citation-Worthiness Focus
I emphasized factual accuracy and clear attribution. LLMs cite sources that provide authoritative, specific information. Generic marketing fluff doesn't get cited - specific, useful insights do.

Step 4: Multi-Modal Integration
I incorporated charts, tables, and structured data alongside text. AI systems can process multiple content formats, and visual elements often get referenced when they clearly illustrate a point.

Step 5: Topical Breadth and Depth
Rather than creating thin content across many topics, I focused on comprehensive coverage of fewer topics. LLMs favor sources that thoroughly address all facets of a subject.

The implementation process involved auditing existing content, identifying which sections were "citation-ready," and systematically restructuring the rest. I created templates that writers could follow to ensure consistency across all new content.

Most importantly, I maintained the traditional SEO foundation. This wasn't about replacing proven strategies - it was about evolving them for a multi-platform search future.

Foundation First

Traditional SEO remains your starting point - build solid content fundamentals before optimizing for AI citation

Content Chunks

Structure each section to stand alone with clear context - LLMs extract passages, not full pages

Citation Signals

Focus on factual accuracy and specific insights rather than generic marketing content

Multi-Format

Integrate charts, tables, and structured data - AI systems process multiple content types for comprehensive answers

The results were encouraging, though I want to be honest about the timeline and expectations. The couple dozen LLM mentions we achieved weren't from aggressive GEO tactics - they came from solid, comprehensive content that naturally aligned with how AI systems process information.

What moved the needle wasn't revolutionary AI optimization techniques. It was focusing on traditional SEO fundamentals first, then adding the layer of chunk-level thinking. The content that got cited by LLMs shared common characteristics: it was authoritative, specific, well-structured, and provided genuine value.

The most cited content types were:

  • How-to guides with clear step-by-step structure - AI loves actionable information

  • Comparison tables and feature breakdowns - Structured data performs well

  • Industry-specific insights with supporting data - Expertise and specificity matter

The timeline was important to understand. Traditional SEO improvements showed up in 2-3 months. LLM citations started appearing around month 4-6, after the content had time to be crawled and integrated into AI training or retrieval systems.

Revenue impact was harder to measure directly since LLM traffic doesn't show up in Google Analytics the same way. But we did see improved brand mentions, more direct traffic, and higher-quality inbound inquiries that suggested people were discovering us through AI-powered research.

Learnings

What I've learned and the mistakes I've made.

Sharing so you don't make them.

After implementing this hybrid approach across multiple client projects, here are the key lessons I learned about optimizing for both traditional search and AI systems:

  1. Don't abandon what works - Traditional SEO fundamentals are still your foundation. LLM optimization is an additional layer, not a replacement.

  2. Think in chunks, not pages - Structure content so any section can stand alone with clear context.

  3. Lead with answers - Put key insights upfront where AI systems can easily extract them.

  4. Specificity beats generality - Generic content doesn't get cited. Specific, actionable insights do.

  5. Multi-format approach works - Combine text with tables, charts, and structured data.

  6. Patience is required - LLM citations take longer to appear than traditional search rankings.

  7. Quality over optimization - Focus on creating genuinely useful content rather than gaming AI systems.

The landscape is evolving too quickly to bet everything on optimization tactics that might be obsolete in six months. The safest approach is building your GEO strategy on top of strong SEO fundamentals, not instead of them.

How you can adapt this to your Business

My playbook, condensed for your use case.

For your SaaS / Startup

For SaaS companies looking to implement this approach:

  • Focus on use-case documentation that can be easily extracted by AI

  • Structure feature pages with clear, standalone benefit statements

  • Create integration guides that work as both user documentation and AI citation sources

  • Build comprehensive comparison content that positions your solution

For your Ecommerce store

For e-commerce stores implementing this strategy:

  • Structure product information for easy AI extraction and comparison

  • Create buying guides that can be cited in AI shopping recommendations

  • Develop category pages with clear, contextual product information

  • Focus on technical SEO fundamentals as your foundation

Get more playbooks like this one in my weekly newsletter