AI Search 2025 Recap: What Worked, What Was Hype, 2026 Plan

Key Takeaways
AI search is becoming a decision layer: people are increasingly using LLMs for high-stakes comparisons at the end of the buying journey, asking which product or vendor to choose when they are close to converting. That makes AI visibility strategically important even when it produces no direct click.
Revenue and conversions matter more than AI visibility alone: revenue and conversions as the primary KPIs. Marketers can observe whether a brand appears in an answer and whether a conversion happens, but the path between those points is often opaque because much of AI search is zero-click.
Original research and unique information are the content advantage: the content that is performing well across AI search is original research, data, UGC, and authentic conversations.
The difference between SEO and AEO has been overstated: The magnitude of the differences between SEO and AEO was overhyped. There are differences, but most SEO tactics keep working, while testing the areas that behave differently in AI search rather than treating AEO as an entirely separate discipline.
AI platforms may move from recommendation to conversion: Expect more native checkouts, native conversions, and LLM experiences inside websites. Agentic commerce and integrations could let transactions happen directly inside AI products rather than only sending users to another site.
Summary
The conversation separates AI-search hype from the areas where the panel is seeing meaningful change. Ethan argues that AI search is still much smaller than traditional search and that the differences between SEO and AEO have been overstated, while the group points to specific shifts that matter: LLMs being used for product decisions, AI answers generating influence without clicks, and new forms of native conversion. They also compare what has worked and what has been overhyped.
Measurement is one of the clearest challenges. Lily prioritizes revenue and conversions, while Ethan describes the journey as measurable at the beginning and end but difficult to observe in the middle. Because users may see a brand in an AI answer and later navigate directly or search for the brand elsewhere, referral traffic alone can understate AI's role in the journey. The panel therefore discusses self-reported attribution, brand demand, and visibility alongside business outcomes.
More content is not automatically better. Lily emphasizes original research, data, UGC, and net-new insight; Ethan calls for rigorous, reproducible experiments rather than unsupported claims. Looking ahead, he expects more native checkouts, native conversions, and AI experiences embedded in websites. Finally they talked about Google and OpenAI being the dominant AI-search platforms in the near term.
I want LLMs to be the closer for me and I want to give them the content to help me close.
Ethan Smith, CEO of Graphite


