Venkata Pagadala, a search intelligence leader at AT&T, argues that zero search volume often indicates high-demand long-tail queries because traditional tools fail to track queries exceeding 3-4 words. He leads search intelligence at enterprise scale and is pursuing a PhD in artificial intelligence, providing him a technical perspective on how LLMs retrieve and prioritize data. In-house SEO managers and content strategists should act on his framework to identify demand that current keyword tools miss.
Who should act on AI Search Optimization and who should skip it
Readers who should act on AI Search Optimization
- In-house SEO Managers should adopt the persona-based query fan out to defend their strategies as search behavior shifts toward AI-driven answers.
- Content Strategists should prioritize specific social platforms like TikTok or Reddit depending on which LLM their target audience uses most frequently.
- SEO Agency Staff should use autonomous AI agents to provide clients with more sophisticated competitive intelligence than standard keyword volume reports.
- Technical SEOs should understand how different LLMs prioritize diverse source signals like YouTube and TikTok to improve retrieval accuracy.
Readers who should skip AI Search Optimization
- PPC Specialists should skip this content and focus on Google Ads or social media ad guides, as this framework focuses on organic AI retrieval.
- Beginner Marketers should skip this and look for basic SEO introductory guides before attempting to build autonomous AI agents or use Open Router.
What to do next about AI Search Optimization
- Map out 3-5 buyer personas and their corresponding product or service offerings to establish the basis for query fan outs.
- Analyze real-time queries on Reddit, Facebook groups, YouTube, and TikTok to identify how people actually describe pain points.
- Create an AI agent in Claude with a single defined objective, such as identifying all speaking opportunities in a specific niche.
- Connect the agent to an API via Open Router to access various AI models for a monthly fee of $10 to $20.
- Store the collected data in a database like Postgres or Railway hosting with timestamps to track how competitors and trends move over time.
User-first intelligence shifts focus from keywords to user journeys
User-first intelligence requires shifting the primary unit of SEO measurement from individual keywords to the entire user journey. Traditional keyword research often fails because it treats a query as an isolated event rather than a step in a process that includes discovery, evaluating, comparing, and deciding. Identifying the gaps in these journeys allows brands to build products that solve specific pain points which are often never captured by standard keyword tools.
Long queries exceeding 3-4 words reveal hidden demand
Zero search volume is not zero demand, as queries that exceed 3-4 words are frequently ignored by traditional tools even though they represent real user intent. For example, a query for a specific vehicle model like the Toyota Grand Highlander may be too long for a tool to track but is a high-intent search for a buyer. SEO practitioners should stop looking at search through an SEO lens and start looking like a user who is searching for a specific product review or comparison.
The Persona-Based Query Fan Out framework identifies high-intent stages
The Persona-Based Query Fan Out is a methodology for generating queries based on specific user personas and their progression through a journey. This method identifies queries at four distinct stages: discovery, evaluating, comparing, and deciding. By mapping these stages, practitioners can find the specific questions users ask when they are ready to buy, which are often more complex and longer than the head terms found in standard tools.
LLM retrieval patterns vary by source platform
Different LLMs have different “biases” for data sources because they are trained to prioritize different data clusters. For instance, Google and Gemini may leverage more YouTube and TikTok content, while ChatGPT often pulls more from Reddit and Facebook groups. Claude is often the default for developers, meaning its specific retrieval patterns should be tested with sample data sets if you are launching a developer-focused product.
LLM Retrieval Source Preferences
- ChatGPT
- Favors Reddit and Facebook groups
- Google / Gemini
- Leverages YouTube and TikTok
- Perplexity
- Pulls from Reddit and Facebook
- Claude
- Preferred by developers; requires specific testing
Autonomous AI agents scale competitive intelligence research
Autonomous AI agents can scale competitive intelligence by monitoring specific niches and identifying new players that keyword tools miss. By defining a single objective for an agent—such as finding all podcast hosts in a specific industry—and connecting it to an API, users can collect and classify data automatically. These agents can track how often a competitor is mentioned or what specific topics they are covering, providing a layer of intelligence that humans cannot replicate at scale.
Execution of AI-driven content is currently easy, but distribution and understanding the audience remain the primary hurdles. Using a platform like Open Router, which was recently acquired by Nvidia for $8 billion, allows users to access various AI APIs for a monthly fee of $10 to $20 to build these systems. This infrastructure allows for the collection of millions of queries and the tracking of engagement metrics over time.
A local construction company identifies high-intent roofing questions
Suppose Sarah, a marketing manager at a local construction company, wants to find out why people choose specific roofers. Sarah identifies the homeowner persona and analyzes Reddit threads for “roofing” and “construction” mentions to find recurring pain points. She then creates an AI agent to monitor for new competitors mentioned in these threads and maps these findings to a “comparing” and “deciding” query fan-out. The company identifies 5 high-intent questions that competitors are not answering, allowing them to create targeted content that captures users at the moment of purchase.
If Sarah had only used traditional keyword tools, she would have seen zero volume for the complex, 4-word questions homeowners actually ask during the comparison phase. By ignoring those long-tail queries, she would have missed the highest-intent demand in the local market.
Google, Anthropic, and SEO research agents lack TikTok source weights, Claude priorities, and costs
- Google has not announced the specific weight or percentage of TikTok sources in its AI Overviews.
- Anthropic has not publicly detailed the specific retrieval priorities for Claude beyond its status as a developer favorite.
- The exact cost of running large-scale autonomous agents for SEO research has not been established as a standard figure.
Frequently Asked Questions
How do you track AI searches?
To track AI searches, identify your buyers, define your product or service offerings, and analyze real-time queries across Reddit, Facebook groups, YouTube, and TikTok. You should then filter these queries to create a persona-based query fan out based on the logic of how people currently interact with these platforms.
Does zero search volume mean there is no demand?
Zero search volume does not mean there is no demand when the query length crosses 3-4 words. Traditional tools often fail to track these longer strings, but they represent the real intent of users who are moving through the discovery or deciding stages of a journey.
Which AI model should I use for developer-focused products?
Developers often expect and use Claude by default, so you should ensure to test Claude’s retrieval specifically for developer-heavy products. Use a sample set of data to see how Claude surfaces your specific product or service compared to other models like ChatGPT.
How do I build an AI agent for SEO research?
To build an AI agent for SEO research, define a single objective such as finding podcast opportunities and follow the official API rules of the chosen model. You can use an aggregator like Open Router to access various APIs and store the resulting data in a database like Postgres to track trends over time.
Source: Venkata Pagadala 📊 How to Find What AI Search Trust (SEO Video Show).