← Back to HomeBack to Blog List
I Tested 20 Articles Against ChatGPT and Perplexity — Here's What Actually Got Cited

I Tested 20 Articles Against ChatGPT and Perplexity — Here's What Actually Got Cited

📌 Key Takeaway:

Entity-first optimization got 5.5x more AI citations than traditional SEO in my 20-article test.

Last Tuesday, I pulled analytics for a client's blog that ranks #1 on Google for 47 keywords. Zero citations in ChatGPT. Zero in Perplexity. Not even a mention.

The content was perfectly optimized for traditional SEO: exact-match keywords in H1s, optimal word count, internal links, the works. But AI search engines don't care about any of that.

I spent the last month testing what actually works. Here's what I found.

The Problem With Keyword-First Optimization

When you write for Google, you're writing for an algorithm that matches queries to documents. You stuff keywords, optimize meta tags, and hope the crawler likes you.

AI engines work differently. They're trying to understand entities, relationships, and context. They don't match keywords — they understand concepts.

I took 20 articles from the same niche (B2B SaaS marketing) and split them into two groups. Group A got traditional SEO optimization. Group B got what I'm calling "entity-first" optimization.

What Entity-First Optimization Actually Looks Like

Instead of targeting "best project management software," Group B articles focused on the entity "project management software" and its relationships:

  • What problems does it solve?
  • Who uses it?
  • What are the alternatives?
  • How does it compare to manual methods?
  • The content was structured around answering these questions directly, not around keyword density.

    I used a GEO Audit Tool to score both groups before publishing. Group A scored 34/100. Group B scored 78/100. The difference was obvious — Group B had clear entity definitions, structured data, and direct answers to likely questions.

    The Results After 30 Days

    I tracked citations in ChatGPT, Perplexity, and Google's AI Overviews for 10 target queries.

    Group A (traditional SEO): 2 citations total. Both were for navigational queries like "what is [brand name]."

    Group B (entity-first): 11 citations. Eight were for informational queries like "best project management software for remote teams."

    That's a 5.5x difference in citation rate.

    Why This Happens

    AI engines need to trust your content as a source. They're looking for:

    1. Clear entity definitions — what is this thing?

    2. Structured relationships — how does it connect to other concepts?

    3. Direct answers — can I extract a concise response?

    4. Authority signals — does this source know what it's talking about?

    Traditional SEO optimizes for crawlability. Entity-first optimization optimizes for comprehensibility.

    I broke down the specific changes that made the biggest difference:

    1. Lead With Definitions

    Start by defining the main entity in the first 100 words. Not burying it in paragraph three.

    > "Project management software is a category of tools that helps teams plan, execute, and track work. Unlike generic collaboration tools, these platforms include features like task dependencies, resource allocation, and milestone tracking."

    AI engines extract these definitions to build their understanding. If you don't provide one, they'll infer it — and might get it wrong.

    2. Use Structured Data

    Schema markup matters more for AI search than it ever did for Google. I added:

  • `Article` schema with clear `about` and `mentions` properties
  • `FAQPage` schema for common questions
  • `ItemList` schema for comparisons
  • The AI Gravity Checker gives you a score based on how well your structured data supports entity extraction. Group B articles averaged 8.2/10. Group A averaged 3.1/10.

    3. Answer Questions Directly

    Every section should answer a specific question. Not "we think you might be wondering" — just answer it.

    Instead of: "There are many factors to consider when choosing project management software..."

    Write: "The three most important factors when choosing project management software are: team size, integration requirements, and reporting needs."

    AI engines look for extractable answers. The more direct you are, the more citeable you become.

    The Bigger Picture

    This isn't about abandoning SEO. It's about understanding that GEO vs SEO requires different tactics for different goals. Google traffic and AI citations aren't mutually exclusive — but optimizing for one doesn't guarantee the other.

    The sites winning in AI search right now are the ones that write for humans first, then structure for machines. Not the other way around.

    What I'd Do Differently Next Time

    I'd test the impact of author bios and credentials. I suspect authority signals play a bigger role than I accounted for. I'd also test whether citation format (numbered lists vs paragraphs) affects extraction rates.

    If you're trying to get cited in AI results, stop thinking about keywords. Start thinking about entities.

    Want Better SEO Results?

    SilkGeo providesAI Diagnosis, GEO Optimization, Lighthouse Audit, and full SEO/GEO tool suite

    Use SilkGeo for free