Perplexity AI Interview: What SEOs Need to Know About AI Search
I recently spoke with Jesse Dwyer of Perplexity about AI search and SEO, focusing on what publishers and SEOs should prioritize to optimize for AI-driven search experiences. His insights reveal how AI search fundamentally differs from traditional SEO and why personal context and sub-document indexing are game-changers.
AI Search Today: Personalization Changes Everything
Dwyer emphasized that AI search is no longer a zero-sum game. Two users entering the same query may receive different answers depending on the AI tool and its use of personal memory.
“Two people with the same query can get a different answer on commercial search if the AI tool they’re using loads personal memory into the context window (Perplexity, ChatGPT). Most traditional SEO best practices still apply, but the results are personalized.”
Key takeaway: SEO still determines whether content is eligible to be retrieved, but visibility is no longer universal. Personal context can significantly change which content a user sees.
Sub-Document Processing: The Future of AI Search
Dwyer highlighted a critical distinction in indexing approaches:
-
Whole-document indexing: Traditional SEO indexes entire pages, scores them, and retrieves top documents. AI tools on this architecture summarize these results (e.g., ChatGPT’s web search). This is the basis of GEO (Generative Engine Optimization).
-
Sub-document indexing: AI indexes granular snippets—small, tokenized fragments—rather than full pages. When a query is made, the AI retrieves tens of thousands of relevant snippets to fill its context window, maximizing accuracy and minimizing hallucinations.
“Instead of retrieving 50 documents, it retrieves about 130,000 tokens of the most relevant snippets… The goal is to completely fill the AI model’s context window with relevant information, leaving little room for it to make things up.”
This approach forms the foundation of AEO (Answer Engine Optimization), where relevance is measured at the snippet level rather than the page level.
Why Personal Context Matters
AI search tools can incorporate user-specific context far beyond what Google profiles provide. This means:
-
Results are personalized for each searcher
-
Two users with identical queries may see completely different underlying sources
-
The AI can reduce bias and prioritize relevant, merit-based answers
Dwyer explained that the competitive advantage for AI search platforms lies in indexing technology and retrieval methods, including:
-
Modulating compute
-
Query reformulation
-
Proprietary models that optimize snippet relevance
Implications for SEOs
While traditional SEO remains relevant for content eligibility, AEO introduces new considerations:
-
Focus on structured, high-quality content fragments that AI can retrieve as relevant snippets
-
Optimize for clarity, context, and relevance, not just keywords or backlinks
-
Understand that AI search may reduce click-through dependency as answers are generated directly in the interface
“The difference between whole-document and sub-document retrieval is night and day… answer quality is constrained by context-window saturation, and accuracy emerges when the model’s window is filled with relevant fragments.”
Takeaways
-
AI search is personalized: Two users may see different results for the same query.
-
Sub-document processing is the future: Optimizing content fragments matters more than traditional page-level SEO.
-
AEO vs. GEO: Whole-document indexing supports GEO; snippet-level indexing drives AEO.
-
Traditional SEO still matters: It ensures content is eligible for retrieval.
-
Accuracy depends on context-window saturation: Filling the AI’s context window with relevant data reduces hallucinations.
For SEOs and publishers, the shift to AI search means rethinking content strategy, indexing, and relevance—moving from page-level optimization to snippet-level precision.