AI & Automation
Personas
SaaS & Startup
Time to ROI
Medium-term (3-6 months)
Last year, I was running a complete SEO overhaul for an e-commerce Shopify client when something unexpected happened. While tracking their traditional search rankings, I discovered their content was appearing in AI-generated responses from ChatGPT and Perplexity—despite being in a niche where LLM usage isn't common.
This wasn't something we initially optimized for. It happened naturally as a byproduct of solid content fundamentals. But it got me thinking: if traditional SEO is about optimizing for Google's algorithm, what does optimization look like for generative AI systems?
That discovery led me down the rabbit hole of GEO (Generative Engine Optimization)—and honestly, most of what the industry is saying about it is wrong. Through conversations with teams at AI-first startups like Profound and Athena, plus my own experiments, I realized everyone is still figuring this out.
Here's what you'll learn from my real-world experience:
Why LLMs were already mentioning my client's content without any "GEO" tactics
The chunk-level thinking approach that actually works for AI indexing
Why traditional SEO foundations matter more than shiny new GEO tactics
The five tactical optimizations I implemented (and their results)
Why you shouldn't abandon traditional SEO for AI optimization
Reality Check
What the AI optimization gurus aren't telling you
If you've been following the latest marketing trends, you've probably heard about GEO (Generative Engine Optimization). The industry is buzzing with "revolutionary" tactics to get your content featured in ChatGPT, Claude, and Perplexity responses.
Here's what most "experts" are recommending:
Keyword stuffing for AI: Cramming content with phrases like "according to experts" and "comprehensive guide"
Citation-baiting: Creating content specifically designed to be quoted by AI systems
Abandoning traditional SEO: Shifting entire strategies to focus on AI optimization
Complex prompt engineering: Writing content that "tricks" LLMs into featuring it
Volume over quality: Pumping out AI-friendly content at scale
This advice exists because everyone's scrambling to get ahead of the curve. AI-generated search is clearly the future, and marketers want to be first-movers. The problem? Most of this advice treats GEO like a completely separate discipline from traditional SEO.
But here's what I discovered through actual client work: the fundamentals haven't changed as much as everyone thinks. LLM robots still need to crawl and index your content. Quality, relevant content remains the cornerstone. Traditional SEO best practices are your starting point, not your replacement.
The industry is making the same mistake it made with voice search optimization—creating an entirely new playbook when the existing one just needs evolution, not revolution.
Consider me as your business complice.
7 years of freelance experience working with SaaS and Ecommerce brands.
So here's the context: I was working with an e-commerce Shopify client who needed a complete SEO overhaul. This was a traditional project focused on improving their Google rankings and organic traffic. Nothing fancy, just solid fundamentals.
The client operated in a pretty niche market—not the type of industry where you'd expect heavy LLM usage. Think specialized B2B products, not trendy consumer goods. We were focused on traditional keyword research, content optimization, and technical SEO improvements.
About three months into the project, I started tracking mentions across different platforms to understand their brand presence. That's when I made an unexpected discovery: their content was appearing in AI-generated responses, despite us never optimizing for it.
We tracked a couple dozen LLM mentions per month across ChatGPT, Perplexity, and Claude. Not massive numbers, but consistent enough to notice. This wasn't something we were tracking initially—it was just a byproduct of our content audit process.
What struck me wasn't just that it was happening, but how it was happening. The content getting picked up wasn't our most SEO-optimized pages. It wasn't the pages with the most backlinks or highest Domain Authority. Instead, it seemed random—or at least, random by traditional SEO standards.
This discovery made me realize that LLMs consume and process content differently than traditional search engines. They weren't just looking at page-level signals; they were breaking content into passages and synthesizing answers from multiple sources.
I started digging deeper, reaching out to teams at AI-first startups to understand what they were seeing. The consensus? Everyone is still figuring this out. There's no definitive playbook yet for GEO, and most of the "tactics" being sold are educated guesses at best.
Here's my playbook
What I ended up doing and the results.
Instead of jumping on the GEO bandwagon with untested tactics, I took a systematic approach. I wanted to understand what was actually working before implementing any changes.
First, I analyzed which of our client's content was getting picked up by LLMs and why. The pattern wasn't immediately obvious, but after several weeks of tracking, I noticed something important: LLMs favored content structured for easy extraction and synthesis.
Traditional SEO optimizes at the page level—title tags, meta descriptions, H1s. But LLMs operate at the chunk level. 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 five-layer approach I developed:
Layer 1: Chunk-Level Retrieval
Instead of optimizing entire pages, I restructured content so each section was self-contained. Every paragraph needed to make sense without the surrounding context. This meant more descriptive subheadings and clearer topic transitions.
Layer 2: Answer Synthesis Readiness
I organized information in logical, sequential structures that LLMs could easily extract and combine with other sources. Think of it as making your content "mix-and-match" friendly for AI systems.
Layer 3: Citation-Worthiness
This wasn't about gaming the system—it was about factual accuracy and clear attribution. LLMs prefer content they can confidently cite, which means being precise with data, sources, and claims.
Layer 4: Topical Breadth and Depth
Instead of targeting single keywords, I covered all facets of topics comprehensively. If someone asked an AI about our client's industry, we wanted to be the go-to source for multiple related subtopics.
Layer 5: Multi-Modal Support
I integrated charts, tables, and visuals not just for human readers, but because LLMs are getting better at processing and describing visual content. This gave us more indexing touchpoints.
The key insight was this: don't abandon traditional SEO fundamentals. Build your GEO strategy on top of strong SEO basics, not instead of them. Every piece of content still needed proper keyword research, technical optimization, and user experience design.
Chunk Thinking
Restructured content into self-contained sections that LLMs could easily extract and synthesize with other sources
Foundation First
Built GEO tactics on top of solid traditional SEO rather than replacing existing optimization strategies
Multi-Modal Approach
Integrated visual content and structured data to give LLMs more ways to process and reference our content
Quality Signals
Focused on factual accuracy and clear attribution rather than trying to game AI algorithms with keyword stuffing
The results weren't dramatic overnight changes—this was a medium-term strategy that compounded over time. Within six months, we saw consistent improvements in AI mentions without sacrificing traditional search performance.
Our LLM mentions increased from a couple dozen per month to consistent visibility across ChatGPT, Perplexity, and Claude responses. More importantly, these mentions were contextually relevant and accurate, not just random citations.
Traditional SEO metrics actually improved alongside our GEO efforts. Conversion rates stayed strong because we hadn't sacrificed user experience for AI optimization. Our approach of building on SEO fundamentals rather than replacing them paid off.
The most interesting result was discovering that content structured for AI consumption was also better for human readers. Chunk-level thinking made our content more scannable and actionable. Self-contained sections improved user engagement metrics.
What surprised me was how natural this evolution felt. It wasn't a radical departure from good content practices—it was more like following the logical next step of content optimization.
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 this real-world GEO experiment:
Foundation matters more than tactics: Solid SEO fundamentals gave us a platform for AI optimization. Without good content basics, no amount of GEO tricks would have worked.
Chunk-level thinking changes everything: Stop optimizing pages; start optimizing passages. Each section should be valuable on its own.
Quality beats gaming: LLMs prefer content they can confidently cite. Focus on accuracy and clear attribution rather than keyword manipulation.
Multi-modal is the future: Visual content and structured data give you more indexing opportunities as AI systems become more sophisticated.
Don't abandon what works: Traditional SEO and GEO can coexist. In fact, they strengthen each other when implemented correctly.
Patience pays off: This isn't a quick-win strategy. AI indexing improvements compound over months, not days.
Context is king: LLMs care more about comprehensive topic coverage than keyword density. Think topics, not just keywords.
The biggest mistake I see businesses making is treating GEO as a replacement for traditional optimization. It's an evolution, not a revolution. The companies that will win in the AI-driven search era are those who build on proven foundations rather than chasing shiny new tactics.
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:
Start with comprehensive SEO foundations before adding GEO tactics
Structure help documentation and feature pages for chunk-level consumption
Create topic clusters covering all aspects of your product category
Integrate customer data and case studies for citation-worthy content
For your Ecommerce store
For e-commerce stores implementing this strategy:
Optimize product descriptions and category pages with comprehensive topic coverage
Create buying guides structured for easy AI extraction and synthesis
Focus on factual product information that LLMs can confidently reference
Build review systems that generate citation-worthy social proof