Optimizing for LLMs, "Dark Social," and Walled Gardens

Optimizing for LLMs, "Dark Social," and Walled Gardens

Kevin Lee interviews John Busby, CMO of Centerfield, to discuss what’s now being called the “new SEO:” how visibility is shifting from classic search results to large language models (LLMs), social platforms, and private communities.

John explains how Centerfield’s comparison properties were built around strong E‑A‑T principles (expertise, authority, trust): real humans testing products, deep reviews, and structured listicle-style content. While traditional organic traffic has declined as users get more answers directly from AI tools, that same content now drives citations within LLM responses. In other words, good SEO content hasn’t stopped working; it’s just shifted upstream into the models that power modern search experiences.

Kevin and John explore the “return of the author.” With AI making content cheap to generate, who is behind an article or review becomes an increasingly important trust signal. John and Kevin discuss how LLMs could start to weigh attributes like: Is this person quoted by major publishers? Do they speak at conferences? Do they publish original research? For brands and publishers, that suggests doubling down on real experts, clear bylines, and thought leadership that builds individual as well as domain authority.

John also highlights the unexpected power of video. Centerfield invested early in YouTube reviews and thought-leadership content, often without a clear ROI story. Today, those videos (and their transcripts) are being consumed by LLMs and helping establish authority for Centerfield and its clients. That combination of authentic human video and searchable transcripts is becoming a significant visibility asset.

We then turn to walled gardens like Google, Meta, and LinkedIn. Their message (“give us your budget and conversion pixel, the AI will do the rest”) is tempting, but dangerous if you’re optimizing only to leads or surface-level metrics. John argues for full-funnel measurement to actual revenue and LTV, plus a clear understanding of incrementality so you’re not overpaying for activity that merely harvests demand.