AI & Automation

My Real Experience: From Traditional SEO to GEO Optimization (And What Actually Moved the Needle)


Personas

SaaS & Startup

Time to ROI

Medium-term (3-6 months)

Last year, I was working with an e-commerce Shopify client who needed a complete SEO overhaul. What started as a traditional SEO project quickly evolved into something more complex when we discovered their content was starting to appear in AI-generated responses - despite being in a niche where LLM usage isn't common.

Here's the thing everyone's getting wrong about LLM indexing: they're treating it like traditional SEO. They're obsessing over "optimizing for ChatGPT" while completely missing the fundamentals. I spent months figuring out what actually makes content discoverable by language models, and the answer isn't what you think.

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 after auditing hundreds of pages and tracking LLM mentions across multiple client projects, I've developed a systematic approach that actually works.

In this playbook, you'll learn:

  • Why traditional SEO audits miss 80% of LLM indexing opportunities

  • The chunk-level thinking framework that determines LLM visibility

  • My 5-step content audit process for AI discoverability

  • Real examples of content that gets cited vs. content that gets ignored

  • How to layer GEO optimization on top of existing SEO without starting over

This isn't about chasing the latest AI trend. It's about understanding how content consumption is fundamentally changing and positioning your content to be found in both traditional search and the emerging AI-driven discovery layer.

Industry Reality

What everyone thinks they know about LLM optimization

The SEO industry has been buzzing about "GEO" (Generative Engine Optimization) for months now. Every agency is suddenly an expert on optimizing for ChatGPT, Claude, and Perplexity. The typical advice sounds sophisticated:

  1. Optimize for conversational queries - Write content that answers questions in natural language

  2. Focus on featured snippets - Since LLMs supposedly pull from Google's featured snippets

  3. Use structured data heavily - Schema markup will help AI understand your content

  4. Write in Q&A format - Make everything easily parseable for AI systems

  5. Target "AI-friendly" keywords - Long-tail, question-based search terms

This conventional wisdom exists because it's an easy mental bridge from traditional SEO. Agencies can repackage their existing services with an "AI optimization" label and charge premium rates. The problem? It's mostly wrong.

LLMs don't consume content the same way search engines do. They don't crawl for fresh content daily, they don't rely on backlinks for authority signals, and they definitely don't prioritize content based on traditional ranking factors. They operate on completely different principles - understanding context, synthesizing information from multiple sources, and generating responses based on training data patterns.

The real challenge isn't optimizing for AI algorithms. It's understanding how to make your content genuinely useful for synthesis and citation in AI-generated responses. That requires a fundamentally different approach to content auditing and optimization.

Who am I

Consider me as your business complice.

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

When I discovered that my client's e-commerce content was getting mentioned in LLM responses, I was honestly surprised. We weren't in a tech-forward industry. We weren't targeting AI-savvy customers. Yet somehow, we were tracking a couple dozen LLM mentions per month - and this was happening naturally, not through any deliberate optimization.

This discovery led me down the rabbit hole of understanding how LLMs actually discover and use content. I started auditing our existing content to figure out what was working and why. The traditional SEO audit tools were useless for this. Ahrefs and SEMrush can tell you about search rankings, but they can't tell you if your content is citation-worthy for AI systems.

I reached out to teams at AI-first companies to understand their approach. What I learned was eye-opening: even companies building AI products don't have this figured out yet. Everyone is experimenting, testing different hypotheses, and trying to understand the patterns.

The breakthrough came when I realized we needed to think about content consumption differently. LLMs don't read pages linearly like humans do. They break content into chunks, analyze each section's usefulness for answering specific queries, and synthesize responses from multiple sources. This meant our content audit needed to evaluate chunk-level value, not just page-level optimization.

I started documenting which pieces of our content were getting cited, how they were being used in AI responses, and what characteristics they shared. The patterns that emerged were completely different from traditional SEO success factors.

My experiments

Here's my playbook

What I ended up doing and the results.

After months of experimentation and analysis, I developed a systematic approach to auditing content for LLM indexing. This isn't about abandoning traditional SEO - it's about layering a new evaluation framework on top of your existing optimization efforts.

Step 1: Chunk-Level Content Analysis

Instead of evaluating entire pages, break your content into logical sections. Each paragraph or subsection needs to be able to stand alone as a valuable piece of information. LLMs often extract specific passages to answer questions, so every chunk should be self-contained and contextually complete.

I audit each section by asking: "If this paragraph was the only thing someone read about this topic, would they understand the concept and be able to act on it?" If not, the chunk needs work.

Step 2: Answer Synthesis Readiness

LLMs excel at synthesizing information from multiple sources. Your content needs to be structured in a way that makes synthesis easy. This means clear logical flow, obvious cause-and-effect relationships, and explicit connections between concepts.

I restructured our content so each section could contribute to different types of AI responses - some sections provide definitions, others offer step-by-step processes, others give examples or case studies. This diversity makes the content more useful for synthesis.

Step 3: Citation-Worthiness Assessment

Not all content is citation-worthy. LLMs tend to reference content that is factual, specific, and authoritative. Vague marketing copy gets ignored. Specific data points, clear methodologies, and concrete examples get cited.

I audited our content for "citation triggers" - specific numbers, unique processes, original research, or distinctive perspectives that an AI would want to reference when answering related questions.

Step 4: Topical Breadth and Depth Evaluation

LLMs value comprehensive coverage of topics. Instead of thin content targeting specific keywords, they prefer resources that cover all facets of a subject. I mapped our content against the full scope of topics in our niche, identifying gaps where we needed deeper coverage.

Step 5: Multi-Modal Content Integration

While LLMs primarily work with text, they increasingly understand context from charts, tables, and structured data. I audited our content for opportunities to add visual elements that enhance understanding and provide additional citation opportunities.

The key insight? Quality content for LLMs isn't just well-written - it's genuinely useful for synthesis and citation. Every piece needs to contribute unique value that an AI system would want to reference when generating responses.

Citation Triggers

Specific data points and unique methodologies that make content reference-worthy

Chunk Architecture

Breaking content into self-contained sections that work independently

Synthesis Value

Ensuring each piece contributes to broader knowledge synthesis

Authority Signals

Building content that LLMs recognize as credible and worth citing

The results weren't immediate, but they were significant. Within three months of implementing our LLM-optimized content audit, we saw our monthly LLM mentions increase from a couple dozen to over 100. More importantly, the quality of mentions improved - we were being cited as authoritative sources rather than just mentioned in passing.

Traditional SEO metrics improved as well. Our content became more comprehensive and better structured, which helped with traditional search rankings. Page engagement increased because the content was more useful and easier to scan.

The unexpected outcome? Our content started performing better across all discovery channels - not just AI systems. When you optimize for synthesis and citation, you naturally create more valuable, comprehensive content that performs well everywhere.

What surprised me most was how this approach affected our content creation process. Instead of targeting individual keywords, we started thinking about comprehensive topic coverage. Instead of optimizing single pages, we built content ecosystems where different pieces supported and referenced each other.

Learnings

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

Sharing so you don't make them.

Here are the key lessons I learned from auditing hundreds of pages for LLM indexing:

  1. Traditional SEO skills are still foundational - LLMs still need to discover your content, which means basic SEO hygiene matters

  2. Chunk-level thinking is essential - Optimize sections, not just pages

  3. Comprehensiveness beats keyword targeting - Cover topics thoroughly rather than targeting specific terms

  4. Citation-worthy content has specific characteristics - Factual, specific, and genuinely useful for synthesis

  5. The landscape is still evolving rapidly - What works today might change in six months

  6. Quality content wins in all channels - Optimizing for LLMs improves performance everywhere

  7. Don't abandon traditional SEO - Layer GEO optimization on top of solid SEO fundamentals

The biggest mistake I see teams making is treating this as an either/or decision. You don't need to choose between traditional SEO and LLM optimization. The best approach builds GEO strategies on top of strong SEO fundamentals.

This approach works best for content-heavy SaaS companies and educational resources. It's less effective for pure product pages or marketing copy. Focus your LLM optimization efforts on your most valuable, educational content first.

How you can adapt this to your Business

My playbook, condensed for your use case.

For your SaaS / Startup

For SaaS startups looking to implement LLM content auditing:

  • Start with your knowledge base and help documentation

  • Focus on educational content over promotional pages

  • Create comprehensive guides rather than thin blog posts

  • Track mentions in AI responses as a new metric

For your Ecommerce store

For e-commerce stores implementing this approach:

  • Optimize category descriptions and buying guides

  • Create comprehensive product comparison content

  • Focus on educational content that supports purchase decisions

  • Build topical authority in your product categories

Get more playbooks like this one in my weekly newsletter