Will Scott

Co-founder & CEO, Search Influence

• First website online mid-90s - trained as an architect
• Coined "Barnacle SEO" in 2008 - optimizing platforms you don't own
• First AI content talk: February 2023
• 30 years building on the web, 20+ focused on SEO

GEO Masterclass:
From Framework to System

Leave with tools you can build this week.

"50% Voice by 2026." We're Here. The Reality:

Six Months Ago This Was a "Future Trends" Slide

GEO in One Sentence

AI Didn't Kill Organic. But It's Reshaping It.

AI creates a new visibility surface. The question isn't whether to care - it's whether you're on it.

You Don't Have to Rank #1 to Be Cited

A destination client was invisible in traditional search. At 2,900 pixels of scroll depth, in an AI overview for a key query, they appeared.

This is the biggest mental model shift in GEO.

Step 1: Query Understanding

AI doesn't match keywords - it interprets intent. "Best Italian for a date night near Times Square" isn't three keywords; it's a context-rich instruction.

Step 2: Retrieval

RAG-powered platforms search the web in real time. Multiple simultaneous sub-queries (query fan-out). Google AI Mode issues hundreds of queries for Deep Search.

Step 3: Synthesis

AI pulls from multiple sources and synthesizes a coherent response. It doesn't copy. It distills.

Step 4: Citation

The final step: attribution. Perplexity cites everything it touches. ChatGPT cites less and less. This creates a platform-specific optimization problem.

Good SEO Is Still Your GEO Foundation

• Technical crawlability - still essential
• Content quality - still the core
• Authority signals - now load-bearing, not decorative
• E-E-A-T - AI reads it the same way Google does

Being Cited Is the New Ranking #1

If you're not in the set AI pulls from, you don't get found. AI systems don't scroll to page 2.

Three Principles Across Every Platform

Pick 2-3 Platforms Based on Your Audience

GEO is platform-specific. Don't try to optimize for all 7 at once. Pick based on where your audience lives and what query types matter most.

Now That You Know How the Engines Work...

...let's talk about what they respond to. AI understands concepts and decomposes complex queries before retrieving anything.

One Query → Multiple Sub-Queries

When a user asks a complex question, AI platforms fan out into multiple sub-queries simultaneously. Google AI Mode issues hundreds of queries for Deep Search.

Write for the Query Cluster

• You're not writing for one query - you're writing for the cluster
• Comprehensive topic coverage beats keyword density every time
• Goal: be the source that answers the most sub-queries in your cluster
• Entity authority amplifies: if AI recognizes you in the space, you get pulled more often

AI Understands Concepts, Not Keywords

"Customer service" and "support" and "help desk" are the same concept to AI. Synonym recognition, related-term clustering, topic modeling - how AI evaluates relevance.

Entities vs. Keywords

Your Brand as a Node

Google's Knowledge Graph and similar entity graphs in AI training data create a map of relationships between entities. If your brand is a node on that map with defined attributes and relationships, AI can find and cite you.

Specific + Verifiable = AI-Citable

Don't just mention "customer service" - ground it with specific, verifiable details AI can work with.

Your GEO Inventory: Entity Mapping

Before you can optimize for entities, you need to know what entities you are. Map your brand at the center with five branches:

Neighborhood Effects for Content

Entity salience: how prominently an entity appears in content.

Co-occurrence: which other entities appear alongside yours. If your content consistently co-occurs with high-authority entities in your space, AI calibrates your entity up.

Schema Markup Is Explicit Entity Definition

You're telling AI systems exactly what your content is, what it says, and how to categorize it. JSON-LD preferred.

Organization Schema: Your Entity Registration Form

Name, URL, logo, description, founding date, social profiles, address, area served - all explicit signals AI uses to construct your entity record.

Every FAQ Block Is a Pre-Formatted AI Answer

FAQ schema directly answers the questions users are asking AI. Every FAQ block is a pre-formatted answer AI can retrieve and cite verbatim.

HowTo Schema: Step-by-Step Content AI Loves

HowTo schema structures exactly the kind of procedural content AI synthesizes well. Use it for installation guides, tutorials, processes - anywhere you have clear sequential steps.

Build It, Then Validate It

• Manual JSON-LD - Maximum control. Best for core entity schema.
• SchemaApp - Visual building with templates. Good for teams.
• SchemaWriter - AI-assisted generation. Refine manually before deploying.

What Goes in Your llms.txt

A plain-text file at your root that tells AI systems what your site is, what it does, and which pages are most important for AI consumption.

• Brief description of your organization and purpose
• Explicitly listed key pages with descriptions (definitive guides, FAQ, products)
• Optional: what you don't want AI citing (legal disclaimers, outdated content)

Structure Is Critical for AI Comprehension

• Clear H2/H3 hierarchy - AI uses headings to understand content sections
• Bullet lists and tables - AI extracts structured data preferentially
• Q&A sections - pre-formatted for direct citation
• Summary sections at the top and bottom of key content

Be the Snippet

Put your core answer in approximately 170-180 characters at the top of the section, then expand with detail below.

Think like a featured snippet writer - because you are now writing for AI synthesis, which works the same way.

AI Can't Watch Videos - It Reads

• Full transcripts are mandatory for video GEO
• Detailed show notes for podcasts
• VideoObject schema with chapters and descriptions
• Alt text that describes what's actually in images, not just keywords
• ImageObject schema for important visuals

Converting Promotional Language to Citable Information

Every piece of "we are the best" language is an AI dead end. Every specific, verifiable fact is a citation opportunity.

What AI-Citable Content Looks Like

Content that AI cites has a recognizable pattern:

• Founding date   •  Customer count   •  Quantified outcome
• Named integrations   •  Geographic specificity   •  Credentials

Getting Retrieved Is Not Enough

AirOps studied what separates pages that get retrieved by ChatGPT from pages that actually get cited. The gap isn't authority or rankings — it's formatting and structure.

What AI Wants to Cite — By Stage

Source: AirOps, "From Retrieved to Cited," April 2026 — 217,508 pages, 25 on-page signals, 7,500 commercial prompts

Authentic > Promotional, Always

Building AI Authority Through Community

• Find the subreddits where your customers ask questions
• Answer questions with specific, verifiable expertise (not sales pitches)
• Cite your own resources naturally when genuinely helpful
• Consistency over time builds community standing that AI recognizes

Knowing How to Use AI to Do The Job Is Good.
Building The System That Does It at Scale Is Great.

Content strategy tells you what to build. Agentic workflows build it for you — at scale, on repeat, while you sleep.

You Type One Question. You Get the Answer.

Claude Code queries your Google Ads API and GSC simultaneously, joins the datasets, identifies the overlap, calculates the wasted spend, and returns a table you can act on.

You never opened a spreadsheet.

Paste a URL. Get 5-Platform Content.

The URL-to-Social pipeline retrieves the article, extracts key themes, writes optimized posts for LinkedIn, Threads, X/Twitter, and Bluesky - each respecting character limits and platform conventions - runs a QA layer that flags AI-sounding language, and queues them for scheduling.

Wake Up Knowing Where You Stand

Every Monday morning, your AI Visibility Measurement system queries ChatGPT, Claude, Perplexity, Google AI Overviews, and Meta AI with your 20 key industry questions. Documents which answers cite your brand. Logs results. Compares to last week. Sends a summary report.

Model Context Protocol

MCP is the standard that lets AI agents interact with external tools, APIs, and data sources. Without MCP, Claude can reason but can't act. With MCP, Claude can query your analytics, write to your CMS, read from your database, and call third-party APIs.

Read more: modelcontextprotocol.io

The Scale Problem Is Real

• Manually monitoring 5 AI platforms across dozens of accounts, every week — not realistic
• Optimizing and publishing to 6 social formats per article — not at scale
• Maintaining entity-rich, frequently-updated content across hundreds of pages by hand — impossible

Claude in Your Terminal

Claude Code is Anthropic's CLI tool that runs Claude in your terminal with access to your file system, can execute code, install packages, and - with the right setup - call APIs.

You describe what you want; Claude figures out the implementation.

Full setup guide: searchengineland.com/claude-code-seo-work-470668

Three Components

One Afternoon, No API Knowledge Required

You tell Claude: "Build me a script that pulls GSC data for the last 30 days and returns a table of queries with impressions, clicks, and average position." Claude writes the script, handles the OAuth flow, and tests it.

Step-by-step: searchengineland.com/claude-code-seo-work-470668

You Publish an Article. Now You Need...

Three Stages

Four Genuinely Different Versions

Automation Doesn't Mean Sloppy Automation

AI-generated content that sounds like AI-generated content trains audiences (and AI systems) to discount it. The QA layer flags words and constructions that signal "written by AI."

The Scale Problem in Numbers

Nobody does this manually. Which means most brands have no idea where they appear in AI responses.

Five Data Points Per Platform + Query

1. Is the brand mentioned?
2. Is it cited with a link?
3. Is the mention positive, neutral, or negative?
4. Who else is mentioned (competitive share of voice)?
5. What source is AI citing when it mentions you?

Logged weekly, trended over time. The competitive SOV data is where the strategy lives.

90 Days of GEO Optimization

Anonymized client example: AI visibility across 5 platforms at baseline vs. after 90 days of GEO optimization. Platform-by-platform scorecard showing mention rate, citation quality, and competitive share of voice.

Real Results, Before Agentic Workflows Existed

These case studies are not hypotheticals. two real GEO optimization projects -- an art & design school and a medical school - before agentic workflows existed. Imagine the results with the tools we just demoed.

What They Did

• Wikipedia entries updated with neutral factual program information
• Faculty engaged authentically in Reddit art education discussions
• Entity mapping across all programs, faculty, alumni, and partnerships
• Content rewritten with specific, verifiable facts (graduation rates, employment outcomes, accreditation)

Not the Most Headcount. The Best Systems.

The teams that will dominate GEO in the next five years are not the ones with the most headcount - they're the ones with the best systems.

You Don't Have to Build Everything from Scratch

• The Claude Code setup took one afternoon (follow the SEL article step-by-step)
• The URL-to-Social pipeline is available as a template
• AI Visibility Measurement can be built with Claude + Python in a weekend

Five Things to Take Away

1. Agentic = AI that plans and executes, not just assists
2. Claude Code + Python = GSC/GA4/Ads analysis without reading API docs
3. URL-to-Social pipeline = 4-platform content distribution in 2 minutes
4. AI Visibility Measurement = know where you stand across 5 platforms weekly
5. The teams that win will run at agent speed. Start building.

You've Got the Framework, the Tactics, the Workflows.

Now let's talk about measuring results, picking the right tools, and where this is heading in the next 12 months.

93% No-Click Doesn't Mean Zero Value

93% of AI search sessions end without a click to any website. This sounds catastrophic - it isn't. It means AI is answering questions your content helped answer.

What This Changes for Your GEO Strategy

1. Your organic GEO investment in ChatGPT now competes with paid placements at the bottom of answers
2. Conversation-contextual targeting means brands optimizing their conversational content get better ad relevance
3. The 7.92 citations/answer average is already low - ads may push organic citations further

You Need Metrics Built for the Citation Economy

• Organic traffic is declining by design - AI answers questions that used to drive clicks
• Ranking position in a world where AI can cite position 47 is a misleading signal
• Traditional metrics were built for the ranking economy; GEO needs new ones

Start Manual, Automate Later

20 target queries × 3 platforms × weekly = 1 hour/week. Move to tools when manual becomes unsustainable.

Who Gets Cited When AI Answers Your Key Queries?

SOV in GEO = your citation count / total brand citations for target queries × 100. Track weekly, report monthly.

Your Server Logs Already Show AI Crawler Behavior

• GPTBot = OpenAI
• ClaudeBot = Anthropic
• PerplexityBot = Perplexity
• BingBot enhanced = Copilot

What GSC Can (and Can't) Tell You About AI Visibility

GSC does not currently separate AI Overview impressions from traditional organic. What it does show: impression spikes without click growth -- the fingerprint of AI Overview inclusion. High impressions, flat clicks = you're being cited but not clicked.

Directional in Quarter One

GEO measurement is directional in quarter one -- you're looking for trends and relative movement, not precise attribution. Track. Trend. Act.

The Market Shifted Since October 2025

New entrants, acquisitions, pricing changes, platform coverage expansions. No single tool does everything; stack accordingly.

What to Check Before You Publish

• Search intent: Does this page match what users -- and AI -- expect for the query? Informational, commercial, or navigational?
• Content types: Do you have the formats AI prefers -- lists, stats, definitions, structured Q&A?
• Citations: Are you citing primary sources? Are authoritative sites citing you?
• People Also Ask: Are you answering the full PAA cluster, not just the primary query?

Who's Beating You in AI Responses, and How?

• Scrunch competitive tracking - Platform-by-platform SOV
• Semrush AI competitor analysis - Integrated with keyword data
• xFunnel - Full-service with analyst support (HubSpot/Monday.com client roster)

Five Questions Before Buying Any GEO Tool

1. Which AI platforms does it cover?
2. How does it validate its data?
3. Does it tell me what to do next, or just what happened?
4. Does it integrate with my existing tools?
5. How quickly does the vendor update when AI platforms change?

You Can Start for $0 Today

• Manual query tracking in a spreadsheet (1 hour/week, scales to 20 queries × 3 platforms)
• GSC for AI Overview impressions (free, first-party, already available)
• Cloudflare free tier for bot identification
• The DIY AI Visibility Measurement from Session 3 (Claude + Python + spreadsheet)

Five-Minute Analysis, Five-Week Improvement Roadmap

• Entity clarity - Is your brand clearly defined?
• Schema coverage - What's present and what's missing?
• Content structure - Heading hierarchy, FAQ sections, structured data
• Authority signals - Citations, E-E-A-T markers, author attribution
• llms.txt - Present and correct?

Free tool at ai-grader.searchinfluence.com

These Are in Production as of March 2026

• API-first content delivery at scale
• AI Mode as default search for millions of users
• Conversation-contextual advertising in ChatGPT
• Agentic task completion in AI assistants

Content Designed for Machine Consumption

Already the primary consumption pattern for high-value, high-frequency content: financial data, real estate listings, product catalogs.

Gemini 2.0, Global, 40+ Languages

Google's AI Mode expanded globally in late 2025. Query fan-out, Deep Search (hundreds of simultaneous queries), multimodal input. For millions of users, this is now the default search experience.

The Next Frontier

AI that books your restaurant reservation, not just recommends one. AI agents that research, compare, and purchase on behalf of users.

Multimodal AI Is Already in Google AI Mode

• "What kind of plant is this?" (camera input)
• "What's on this menu?" (photo)
• "Find restaurants near this landmark" (image + location)

Systems, Not Headcount

Not the ones with the most headcount. Not the ones with the best prompt engineers. The ones with the best systems - measurement, content production, entity maintenance, visibility tracking - running at scale.

Five Things to Remember

1. GEO layers on top of SEO -- good SEO is still the foundation.
2. Platform-specific: Perplexity rewards citation volume; ChatGPT rewards authority. One size doesn't fit all.
3. Agentic workflows are available now -- start with one, run it for a month, then build the next.
4. Measurement is directional in quarter one -- track, trend, act. Don't wait for perfect data.
5. The shift from "tools that help you" to "systems that do it for you" is happening now. Get ahead of it.