219: YOUR HOMEPAGE IS NOT YOUR HOMEPAGE ANYMORE WITH RAND FISHKIN
RAND FISHKIN
SparkToro
Co-founder of SparkToro and Moz, author of 'Lost and Founder,' and search marketing veteran researching AI's impact on brand visibility.
Every AI visibility tool promises to track your brand's position in ChatGPT and Perplexity, but there's a fundamental problem: AI rankings don't actually exist. New research reveals that you'd need to ask an LLM for brand recommendations 1,500 times before getting the same list in the same order twice. The entire AI tracking industry is built on a metric that's statistically meaningless.
KEY TAKEAWAYS
- Ask AI tools multiple times (10-12 minimum) before trusting any brand or product recommendation, then compile the full list across all responses
- Stop relying on AI visibility tools that use synthetic queries with forced list lengths, as these don't reflect how real users prompt AI systems
- Focus on generating genuine brand mentions across the web rather than optimizing for AI rankings that statistically don't exist
- Measure visibility percentage (how often you appear) rather than ranking position, since position varies wildly between queries
- Treat every platform that discusses your brand as your homepage, because that's increasingly where first impressions happen
SHOW NOTES
The 1,500 Query Problem
AI tools produce probabilistic outputs, not deterministic answers. When someone asks ChatGPT or Claude for brand recommendations, they're not getting "the answer." They're getting one of thousands or potentially millions of possible answers. The next person asking the same question gets a different list, different order, different number of items. Research across multiple LLMs found that matching two identical brand lists in identical order requires approximately 1,500 attempts.
Why AI Tracking Tools Mislead
Most AI visibility platforms use synthetic queries, prompts generated by asking AI what humans would ask. The problem? When real humans write prompts for the same intent, lexical similarity scores around 0.07, roughly equivalent to the similarity between kung pao chicken and peanut butter. They share a few core ingredients but are fundamentally different.
The tracking tools compound this by forcing specific list lengths: "Give me 10 brands for this category." Real users never prompt this way. They ask naturally, and the AI decides whether to return three brands or fifteen. A brand appearing at position four in a synthetic ten-item list might not appear at all in organic queries returning only three results.
Here's what's strange though. Despite completely different phrasing, the brands produced from human prompts and AI-generated synthetic prompts show significant overlap. The AI understands intent well enough to surface similar brands regardless of how the question is worded. But what happens when visibility tools report a ranking that only exists in their artificial testing environment?
The Only Metric Worth Tracking
Visibility percentage tells you something real: what portion of queries for your category include your brand in the response. Position is noise. Presence is signal. If your brand appears in 60% of responses about your category, that's actionable data. Ranking third versus fifth in any single response is statistical randomness.
Brand Mentions Drive Everything
The best brands of the last three decades weren't built on digital marketing tricks. Nike, Apple, Tesla, none of them grew primarily through SEO or paid search optimization. They built brands that people talk about, recommend, and mention across every platform. Those mentions become training data. They appear in retrieval systems. They shape what AI tools surface.
This principle hasn't changed in the AI era. Getting mentioned on Reddit, YouTube, industry publications, and customer reviews matters more than any prompt engineering strategy. The brands appearing consistently in AI responses earned that visibility through years of building genuine presence across the web.
Your New Homepage
Traditional thinking puts your website at the center of your digital presence. That mental model is outdated. Your homepage is now whatever Google says about you, what ChatGPT summarizes, what Reddit threads discuss, what YouTube reviewers conclude. First impressions happen on platforms you don't control.
This doesn't diminish the importance of SEO. Quite the opposite. SEO careers matter more than ever, but the metric changed. Traffic as the primary goal is dying. Visibility, brand presence, and mentions across platforms where content is consumed without clicks, that's the new game. The skills translate directly. The scoreboard just looks different.
WATCH ON YOUTUBE
QUESTIONS ANSWERED
How consistent are AI brand recommendations?
AI brand recommendations are highly inconsistent due to probabilistic output models. Research shows you would need to ask ChatGPT or Claude approximately 1,500 times to receive the same list of brands in the same order twice. Each query produces different brands, different list lengths, and different ordering.
Are AI visibility tracking tools accurate?
Most AI visibility tools provide misleading data because they use synthetic queries that don't match how real users prompt AI systems. These tools often force specific list lengths and use AI-generated prompts that have very low similarity to actual human queries. The visibility percentages they report may not reflect real-world brand presence.
Is ChatGPT bigger than Google search?
No, Google remains approximately 210 times larger than ChatGPT in terms of search usage. While AI adoption has grown significantly with 30-40% of Americans using AI tools regularly, Google's usage has not declined during this period. Google remains the primary search tool for most users.
What metrics should I track for AI visibility?
Focus on visibility percentage rather than ranking position. Visibility percentage measures how often your brand appears in AI responses for relevant queries, regardless of position. Since ranking position varies wildly between queries due to probabilistic outputs, tracking specific rankings provides little actionable insight.
How do brands appear in AI responses?
Brands appear in AI responses through two pathways: base model training and retrieval augmented generation (RAG). Base model visibility comes from brand mentions in training data collected from across the web. RAG pulls real-time information from indexed sources. Both pathways favor brands with strong, consistent mentions across multiple platforms.
What is zero-click marketing?
Zero-click marketing shifts focus from driving website traffic to building brand visibility on platforms where content is consumed without clicks. As AI overviews and social platforms increasingly answer questions directly, the goal becomes ensuring your brand is mentioned and visible wherever potential customers encounter information about your category.
RELATED ARTICLES
Browse all articlesENJOYING THIS EPISODE?
No Hacks explores how to optimize websites for AI agents, with weekly episodes featuring SEOs, developers, and AI researchers. Subscribe on your favorite platform.
Subscribe Now