Upgrade to Pro — share decks privately, control downloads, hide ads and more …

One size doesn’t fit all: optimizing for AI sea...

One size doesn’t fit all: optimizing for AI search across different LLMs

Presented @ BrightonSEO San Diego in 2025.

Here’s the harsh truth: There is no universal AI search optimization strategy.

Why? Because different large language models (LLMs) — like Google AI Overviews, ChatGPT, and Perplexity — are trained on different data, use different methodologies, and generate results in different ways.

If you’re treating them all the same, you’re leaving visibility (and results) on the table.

In this session, we’ll break down:

• How major AI models differ in training data, ranking factors, and content retrieval.

• How to tailor your content to increase the chances of being cited, summarized, and surfaced across different AI-driven platforms.

• Actionable GEO (generative engine optimization) strategies for Google, OpenAI, and emerging AI engines.

• What brands and marketers need to do now to avoid becoming irrelevant in an AI-first search landscape.

Avatar for Sam Richardson

Sam Richardson

September 22, 2025
Tweet

More Decks by Sam Richardson

Other Decks in Marketing & SEO

Transcript

  1. Optimizing for AI Search Across LLMs Sam Richardson Intero Digital

    speakerdeck.com/srichardson /samuelwrichardson One Size Doesn’t Fit All:
  2. Sam is the vice president of growth at Intero Digital,

    a digital marketing agency with an emphasis on SEO and paid search.
  3. Layer Purpose RAG Grounds the LLM's responses in fresh external

    data to increase accuracy and relevance
  4. | How Engines Learn v Entity Associations Models learn which

    concepts and entities connect Context and Co-occurrence Meaning is shaped by how often and where they appear together Probabilistic Patterns Co-occurrences form statistical patterns captured in embeddings
  5. Trained on high-trust sources available during training - public corpora,

    archived web Pre-Trained LLM Fixed training data
  6. Consistent brand and entity mentions Evergreen, entity-rich content Mentions in

    high-trust, longstanding sources How to Influence Pre-Trained LLM Fixed training data
  7. Information comes from live external data sources: web indexes, vector

    databases, knowledge graphs. RAG RAG Search-augmented, real-time retrieval
  8. Build topical authority, coverage & depth Structured content formats and

    schema Entity-level trust and authority How to Influence RAG RAG Search-augmented, real-time retrieval
  9. Multiple live sources (web search APIs, verticals, structured data, file

    context) Agentic Capabilities Plans, reasons, and uses tools in multi-step loops
  10. Tool-ready data via structured access Build multi-step contextual relevance Establish

    authority signals for agent trust How to Influence Agentic Capabilities Plans, reasons, and uses tools in multi-step loops
  11. | The Retrieval Framework Presence Recognition Accessibility Are you present

    in AI’s training and retrieval sources? Are you recognized as a relevant and trusted entity? Are your content and the info about your brand accessible to AI?
  12. 5,665% increase in generative search visibility 189% YoY revenue growth

    attributed to AI-driven referrals 96.15% increase in conversion rate from AI-powered sessions
  13. Conduct Entity Research and Identify Semantic Gaps Understand which entities

    define your space and where you're missing from the conversation. 2
  14. Inject Your Brand Into Real-Time Signals Influence the layer that

    relies on fresh data (search-augmented). 5
  15. Sam Richardson Vice President of Growth, Intero Digital [email protected] interodigital.com

    Visit Me at Booth #25 Download the Retrievability Framework