Rand Fishkin ran 2,961 prompts across ChatGPT, Claude, and Google AI.

Fewer than 1 in 100 produced the same brand list. In this conversation, he breaks down why the entire "AI ranking" category is built on probability, what actually moves your brand into LLM answers, and how he's marketing his new product AlertMouse with zero AI search visibility on day one.

Rand Fishkin is the cofounder and CEO of three companies: SparkToro, software that makes audience research accessible to everyone, indie game maker Snackbar Studio, and the the superior alternative to Google Alerts: Alertmouse.

What you'll learn:

→ Why AI answers are probabilistic, not deterministic, and what to track instead of rankings

→ The two real mechanisms that influence LLM visibility (RAG and training data)

→ How user prompts actually behave when 150 people are asked the same question

→ Why PR and positioning are now 100x more important than tactical SEO

→ The white hat playbook for showing up in ChatGPT, Gemini, and Google AI Overviews

→ How to use AlertMouse and SparkToro together to find prompts your buyers actually use

→ What "audience intelligence" vs. "audience research" teaches you about brand positioning lag

→ The Seattle Ultrasonics knife story: how one founder turned mentions into product roadmap

→ Where Rand says marketers should spend attention in 2026 (and what to ignore)

Chapters:

0:02 Intro

0:38 The Datos prompt study and why AI answers are random

4:13 The two mechanisms that move LLM visibility

8:47 How Rand is marketing AlertMouse from zero visibility

13:42 Why positioning is the biggest AEO lever

17:08 AlertMouse use cases and the Seattle Ultrasonics story

22:11 Where to actually invest attention in 2026

Connect with Rand:

SparkToro: https://sparktoro.com

AlertMouse: https://alertmouse.com

LinkedIn: https://www.linkedin.com/in/randfishkinWhat are other ideas that we can add for the YouTube thumbnail? I don't really love "AI rankings don't exist." What are some of the main takeaways from the episode?