AI & Automation
Personas
SaaS & Startup
Time to ROI
Medium-term (3-6 months)
Last year, I worked 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.
Even in a traditional e-commerce niche, we tracked a couple dozen LLM mentions per month. This wasn't something we initially optimized for - it happened naturally as a byproduct of solid content fundamentals. This discovery led me down the rabbit hole of GEO (Generative Engine Optimization).
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 I've learned some valuable lessons about how AI systems like Perplexity actually consume and present content.
What you'll learn from my experience:
How Perplexity and other AI engines actually process metadata
The real difference between traditional SEO signals and GEO optimization
Why chunk-level thinking matters more than page-level optimization
A practical framework for adapting your content strategy
What actually moved the needle in our AI content optimization experiments
Conventional wisdom
What every SEO expert is telling you
Most SEO professionals are approaching AI optimization the same way they've always approached search: focus on metadata, keywords, and traditional ranking factors. The conventional wisdom suggests that AI systems like Perplexity work similarly to Google - crawling pages, reading meta descriptions, and using title tags to understand content.
Here's what the industry typically recommends for AI optimization:
Optimize meta descriptions - The belief that AI systems read and prioritize meta descriptions for content summaries
Perfect your title tags - Assuming AI engines use H1 tags and titles as primary content signals
Traditional keyword density - Applying old-school keyword optimization techniques
Schema markup focus - Believing structured data is the key to AI visibility
Page-level optimization - Treating each page as an isolated ranking unit
This conventional approach exists because it's comfortable. It's what we know. SEO professionals are applying familiar frameworks to unfamiliar technology. The problem? AI systems don't consume content the way traditional search engines do.
Where this falls short in practice is fundamental: AI engines like Perplexity break content into passages and synthesize answers from multiple sources. They don't just read your meta description and decide whether to rank your page. They're looking for specific, factual content that can be extracted and combined with information from other sources.
The traditional page-centric approach misses how these systems actually work at the content consumption level.
Consider me as your business complice.
7 years of freelance experience working with SaaS and Ecommerce brands.
When I started working with this e-commerce client, I had no intention of diving into AI optimization. We were focused on traditional SEO fundamentals - fixing technical issues, improving page speed, and creating comprehensive product content.
The client operated in a niche market selling specialized equipment. Not exactly the type of industry where you'd expect AI adoption, right? But three months into our SEO overhaul, something interesting happened.
During a routine competitor analysis, I decided to test how our content was performing across different platforms. On a whim, I asked Perplexity a question related to our client's main product category. To my surprise, their content appeared in the AI-generated response - not prominently, but it was there.
This discovery was accidental, but it sparked my curiosity. I started tracking mentions across different AI platforms systematically. What I found was fascinating: despite being in a traditional industry, our client was getting mentioned in AI responses about two dozen times per month. This wasn't from any special optimization - it was happening organically.
The breakthrough moment came when I realized these mentions weren't correlating with our highest-ranking pages on Google. The content that AI systems were pulling from was often buried deep in our site - product specifications, technical documentation, even FAQ sections that barely got organic traffic.
This pattern completely challenged my assumptions about how AI systems consume content. It wasn't about page authority or traditional ranking signals. Something else was happening.
Here's my playbook
What I ended up doing and the results.
Once I realized that AI systems were selecting content differently than search engines, I developed a systematic approach to understand what actually influences AI content selection. This wasn't about replacing traditional SEO - it was about building an additional layer on top of solid fundamentals.
The Foundation: Chunk-Level Content Structure
The first major insight was that AI systems like Perplexity don't consume pages - they consume passages. Each section of content needs to be self-contained and valuable on its own. I restructured our content so that every heading and subheading could stand alone as a complete answer to a specific question.
Instead of writing flowing narrative content, I broke everything into discrete, factual chunks. For example, rather than "Our products are made with high-quality materials that ensure durability," I rewrote it as "These products use 316-grade stainless steel construction, which provides corrosion resistance for 15+ years in marine environments."
Metadata Reality Check
Here's where it gets interesting: traditional metadata barely mattered. I ran experiments comparing pages with perfect meta descriptions versus pages with minimal metadata. The AI mention rate showed no significant correlation with meta description quality.
What did matter was the actual content structure within the page. AI systems seemed to prioritize factual statements, specifications, and concrete data points over marketing copy or emotional appeals.
The Citation-Worthiness Factor
I noticed that content getting mentioned by AI systems shared specific characteristics: clear attribution, factual accuracy, and data that could be easily extracted and verified. I started optimizing for "citation-worthiness" rather than "keyword density."
This meant including specific numbers, dates, technical specifications, and industry standards. Instead of saying "fast shipping," I wrote "ships within 24 hours via UPS Ground to continental US addresses."
Multi-Modal Content Integration
AI systems excel at processing different content types simultaneously. I integrated charts, tables, and visual specifications directly into our text content. This wasn't just for user experience - it was feeding AI systems the structured data they prefer in a format they can easily parse.
Technical Insights
How AI systems actually read your content differently than Google's crawlers
Data Structure
Why factual precision beats keyword optimization in AI content selection
Content Chunking
Breaking information into self-contained, citation-worthy passages
Attribution Signals
Making your content trustworthy enough for AI systems to reference
After six months of systematic optimization for AI visibility, we saw significant changes in how our content was being discovered and referenced. The couple dozen monthly mentions grew to over 100 references across various AI platforms.
More importantly, these AI mentions started driving qualified traffic. People who discovered our client through AI-generated responses were more likely to convert because they arrived with specific, technical questions that matched exactly what we were selling.
The traditional SEO metrics also improved. Google seemed to favor the more structured, factual content approach. Our average position improved across target keywords, and we saw a 40% increase in organic click-through rates.
The unexpected outcome? Customer support inquiries became more sophisticated. Instead of basic "do you sell X?" questions, prospects arrived asking technical specifications and implementation details. The AI-optimized content had pre-qualified leads better than any traditional funnel.
What I've learned and the mistakes I've made.
Sharing so you don't make them.
The biggest lesson from this experience was that GEO (Generative Engine Optimization) isn't about replacing SEO fundamentals - it's about building on top of them. The content that performed well in AI systems was still well-structured, technically sound, and user-focused.
Chunk-level thinking is essential: Every paragraph should be valuable in isolation
Factual precision beats marketing fluff: AI systems prefer specific, verifiable information
Traditional metadata has limited impact: Focus on content structure over meta descriptions
Citation-worthiness is the new ranking factor: Make your content trustworthy and attributable
Multi-modal integration matters: Combine text, data, and visual elements
Industry doesn't matter: Even traditional sectors can benefit from AI optimization
Quality leads over quantity: AI-discovered prospects arrive more qualified
What I'd do differently: Start tracking AI mentions from day one. I wish I had baseline data before beginning optimization to measure the true impact of our changes.
How you can adapt this to your Business
My playbook, condensed for your use case.
For your SaaS / Startup
For SaaS companies looking to optimize for AI discovery:
Structure feature documentation as self-contained factual blocks
Include specific technical specifications and API details
Create comparison tables with concrete data points
Focus on use case scenarios with measurable outcomes
For your Ecommerce store
For e-commerce stores wanting to improve AI visibility:
Add detailed product specifications with exact measurements
Include material composition and manufacturing details
Create buying guides with specific criteria and comparisons
Structure shipping and return policies as factual statements