Five Percent — Video
Five Percent — Graphite
Video

The ROI of Improving AI Visibility

22.2K views

Key Takeaways

  • Citation overlap reveals how AI search groups related queries: Different prompts that look similar from a keyword perspective can produce different citation sets. Comparing those sources provides a stronger signal for whether AI search treats two queries as the same content opportunity or distinct ones.

  • AI visibility is shaped by the sources behind the answer, not just the brand’s own content: For broad, competitive queries, brands can appear because they are mentioned on third-party sources that AI systems consistently retrieve. Which sources matter varies by query and category, with Reddit, YouTube, affiliates, listicles, and G2.

  • The AI long tail is defined by combinations of user requirements: AI search can surface questions combining features, use cases, integrations, attributes, and other criteria that may have virtually no conventional search volume. The Rippling, and Gusto examples show how these detailed questions can influence product recommendations, as well as showing how missing answers leave a gap competitors can fill.

  • AI citation visibility is probabilistic rather than fixed: The same prompt can produce different sources across repeated runs, with the speakers noting that platforms vary in stability. Asking the same question multiple times can reveal which citations consistently appear and which are more volatile.

  • Consensus answers and Integration content can turn product relationships into visibility opportunities: The Otter/Zoom example demonstrates how integration pages can create a two-sided citation opportunity, allowing a smaller company to build content around its relationship with a larger platform and potentially earn coverage from that partner.

Summary

The discussion explores how citation analysis can reveal which sources influence AI-generated answers and help marketers decide whether different prompts belong on the same URL or require separate pages. Smith also explains how broader queries require visibility beyond a brand’s own website, with sources such as Reddit, YouTube, listicles, affiliates, and partner sites playing a role.

The speakers then examine the expanding AI search long tail, where users ask highly specific questions about features, use cases, integrations, and attributes. They highlight product pages, documentation, help centers, feature pages, and integration content as ways to address these questions, and show how consensus building can help smaller brands get mentioned

The conversation also covers AI result volatility, experimentation and reproducibility, brand reputation, attribution, and how to make content easy for AI systems to retrieve and understand. The overarching message is that marketers can build on existing SEO practices while adapting their approach to the way AI systems retrieve, synthesize, and cite information.

I think people spend way too little on analysis and experimentation design, which is why so much of the work that they do doesn't work.

Ethan Smith, CEO of Graphite