AI & Automation

From Traditional SEO to GEO: My Real-World Experience Optimizing for AI Assistants


Personas

SaaS & Startup

Time to ROI

Medium-term (3-6 months)

Last year, I was working with an e-commerce Shopify client on what seemed like a straightforward SEO project. Traditional keyword research, content optimization, the usual playbook. But something unexpected happened that changed how I think about search entirely.

We started tracking mentions in ChatGPT, Claude, and Perplexity—even though this was a traditional e-commerce niche where LLM usage wasn't supposed to be common. To our surprise, we were getting mentioned in AI responses a couple dozen times per month. This wasn't something we optimized for; it happened naturally as a byproduct of solid content fundamentals.

That discovery led me down the rabbit hole of what's now called GEO (Generative Engine Optimization). While everyone debates whether SEO is dead because of AI, I've been quietly testing what actually works in this new landscape. Spoiler: traditional SEO isn't dead, but it's definitely evolving.

Here's what you'll learn from my hands-on experience:

  • Why traditional SEO still matters as your foundation (and always will)

  • How I discovered content was appearing in AI responses without optimization

  • The chunk-level optimization strategy that increased our LLM mentions

  • Practical GEO tactics you can implement alongside traditional SEO

  • Why betting everything on AI optimization is a mistake

Don't abandon what works. Build on it. Here's how I learned to optimize for both search engines and AI assistants without sacrificing performance in either.

Industry Reality

What the experts keep saying about AI killing SEO

If you've been following SEO discussions lately, you've probably heard the same narrative over and over. "SEO is dead because of AI." ChatGPT is going to replace Google. Everyone will just ask AI assistants for answers instead of searching. Traditional optimization is becoming obsolete.

The SEO industry has split into two camps. On one side, you have the traditionalists insisting nothing has changed and doubling down on the same old tactics. On the other side, you have the "AI-first" crowd claiming you need to abandon everything and focus solely on optimizing for large language models.

Here's what the conventional wisdom says you should do:

  1. Focus entirely on conversational queries - Everything needs to sound like someone talking to ChatGPT

  2. Optimize for featured snippets only - Since AI pulls from these, they're all that matters now

  3. Write for AI consumption - Structure everything for easy extraction by language models

  4. Ignore traditional ranking factors - Page speed, backlinks, and technical SEO are supposedly irrelevant

  5. Create only FAQ-style content - Because that's what AI assistants prefer

The problem with this thinking? It's based on speculation rather than real-world testing. Most people making these claims haven't actually optimized content for AI responses and measured the results. They're reacting to fear rather than data.

What I discovered through actual testing is that the relationship between traditional SEO and AI optimization is much more nuanced than the industry wants to admit.

Who am I

Consider me as your business complice.

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

When I started that Shopify project, I wasn't thinking about AI at all. This was traditional e-commerce SEO—product descriptions, category pages, blog content targeting commercial keywords. Standard stuff that's worked for years.

But I've always been curious about new trends, so I started tracking whether our content was appearing in AI responses. I set up monitoring for mentions across ChatGPT, Claude, and Perplexity, even though our niche (traditional e-commerce products) wasn't typically associated with AI usage.

The surprise came about three months in. We were getting mentioned in LLM responses consistently—around 20-30 times per month. Not massive numbers, but significant for a traditional business that hadn't optimized for AI at all.

Here's what was fascinating: the content that AI assistants were citing wasn't our "best" SEO content by traditional metrics. It wasn't the pages with the highest Domain Rating or the most backlinks. Instead, AI was pulling from pages that had clear, comprehensive information structured in a logical way.

I started digging deeper. Through conversations with teams at AI-first startups like Profound and Athena, I realized everyone was figuring this out in real-time. There's no definitive playbook yet, which means there's a huge opportunity for businesses willing to experiment.

The breakthrough moment came when I realized that LLMs don't consume pages like traditional search engines. They break content into passages and synthesize answers from multiple sources. This meant our content structure needed to work on two levels: page-level for search engines and chunk-level for AI systems.

That's when I developed what I now call a layered approach: solid traditional SEO as the foundation, with GEO techniques built on top. Instead of choosing between optimizing for Google or AI, I found ways to do both effectively.

My experiments

Here's my playbook

What I ended up doing and the results.

After discovering that our content was naturally appearing in AI responses, I developed a systematic approach to optimize for both traditional search engines and AI assistants. This isn't about abandoning SEO—it's about enhancing it.

Foundation Layer: Traditional SEO Still Matters

First, I made sure we had rock-solid traditional SEO fundamentals. LLM robots still need to crawl and index your content, just like Google's bots. Quality, relevant content remains the cornerstone. Technical SEO, page speed, and proper site structure are still critical.

For the e-commerce client, this meant:

  • Comprehensive keyword research targeting commercial intent

  • Optimized product descriptions and category pages

  • Technical SEO audit and fixes

  • Internal linking strategy for crawlability

Enhancement Layer: Chunk-Level Optimization

This is where it gets interesting. I restructured content so each section could stand alone as a valuable snippet. Instead of writing traditional long-form articles, I created content where every paragraph provided complete, useful information.

The five key optimizations I implemented:

  1. Chunk-level retrieval - Made each section self-contained with complete context

  2. Answer synthesis readiness - Structured information in logical, extractable formats

  3. Citation-worthiness - Ensured factual accuracy and clear attribution

  4. Topical breadth and depth - Covered all facets of topics comprehensively

  5. Multi-modal support - Integrated charts, tables, and visuals that AI could reference

Testing and Measurement

I set up systematic tracking across multiple AI platforms. Using custom prompts, I tested how often our content was being cited, in what contexts, and for which types of queries. This data informed our content strategy going forward.

What surprised me most? 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.

The key insight: don't abandon what works. Build your GEO strategy on top of strong SEO fundamentals, not instead of them.

Testing Framework

I developed systematic tracking across ChatGPT, Claude, and Perplexity to measure real AI mention performance, not just speculation about what might work.

Content Structure

Instead of traditional long-form articles, I restructured content so each section could stand alone as valuable, extractable information for AI systems.

Dual Optimization

Rather than choosing between SEO and GEO, I found ways to optimize content that performs well in both traditional search and AI responses simultaneously.

Foundation First

Traditional SEO fundamentals remain critical—LLM robots still need to crawl and index content just like Google's bots, making technical SEO essential.

The results from this layered approach were encouraging, though I want to be realistic about the scale. Our traditional SEO metrics continued to improve steadily—organic traffic, keyword rankings, and conversion rates all moved in the right direction.

But the AI optimization layer added something new. Our LLM mentions increased from occasional to consistent, growing from around 10 per month to 30+ per month over six months. More importantly, the quality of these mentions improved—AI assistants were citing our content for more complex, high-value queries.

The unexpected benefit? Our traditional SEO performance actually improved. Content optimized for chunk-level retrieval performed better in featured snippets and "People Also Ask" sections. The clearer structure and comprehensive coverage that worked for AI also worked for Google's algorithms.

Timeline-wise, traditional SEO improvements showed up within 2-3 months, while consistent AI mentions took 4-6 months to establish. The lag makes sense—AI training data has different refresh cycles than search index updates.

What I learned: this isn't about replacing traditional SEO with something new. It's about enhancing what already works with additional optimization layers that prepare your content for the future of search.

Learnings

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

Sharing so you don't make them.

After a year of testing both traditional SEO and GEO optimization, here are the most important lessons I've learned:

  1. Foundation beats innovation every time - Solid traditional SEO fundamentals outperformed clever GEO tactics. Get the basics right first.

  2. The landscape is evolving too quickly for absolutes - Don't bet everything on optimization tactics that might be obsolete in six months. Diversify your approach.

  3. AI systems favor comprehensive, factual content - The same qualities that make content valuable to humans make it valuable to AI. Focus on quality over optimization tricks.

  4. Chunk-level thinking improves all content - Structuring information for easy extraction benefits both AI systems and human readers.

  5. Traditional metrics still matter - Domain authority, backlinks, and technical SEO continue to influence both search rankings and AI citation likelihood.

  6. Testing beats speculation - The industry is full of theories about AI and SEO. Actual testing reveals what works in practice.

  7. Integration wins over replacement - The most successful approach combines traditional SEO with GEO techniques rather than choosing one or the other.

What I'd do differently: I wish I'd started tracking AI mentions earlier in the traditional SEO process. The insights would have informed our content strategy from the beginning rather than being a later addition.

When this approach works best: For businesses with substantial content needs and long-term growth goals. When it doesn't: If you're looking for quick wins or have very limited content resources.

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 this dual approach:

  • Start with traditional SEO fundamentals for product and feature pages

  • Structure help documentation for both users and AI consumption

  • Create use-case content that works as standalone chunks

  • Track mentions in AI responses related to your industry keywords

For your Ecommerce store

For e-commerce stores adapting to the AI landscape:

  • Optimize product descriptions for both search engines and AI discovery

  • Create buying guides structured for easy AI extraction

  • Focus on factual, comprehensive product information

  • Monitor AI mentions for brand and product-related queries

Get more playbooks like this one in my weekly newsletter