We explore how AI search has become a recommendation engine that reshapes discovery, trust, and buying decisions, and why this creates a rare window for small brands to break through. Chris Donnelly, co-founder of Searchable, shares practical steps to surface in LLM results using structure, context, and community-driven momentum.
• LLMs shifting search from links to recommendations
• Why context and structured data beat keyword stuffing
• How small brands win niche conversational queries
• Practical llm.txt and schema actions that move metrics
• Social content influencing AI answers via LinkedIn and Reddit
• Personal brand as a compounding growth channel
• Platform nuances across ChatGPT, Gemini, Claude, Perplexity
• Reverse-engineering ranking signals through aggregated changes
• AI as augmentor for research, drafts, and prioritization
• A clear roadmap to measure impact and iterate
From Keywords To Conversations: How LLMs Decide What People Buy
AI search has shifted from blue links to guided recommendations, turning language models into the first touchpoint for product discovery. That evolution changes how brands are found, trusted, and chosen, especially when a single conversational query narrows a world of options into five tailored picks. The conversation between Joeri Billast and Chris Donnelly explores how marketers can adapt fast by reframing SEO around context, freshness, and structured data, rather than just keywords. When users ask for “soundproof headphones for a noisy shared office,” they expect an answer with reasoning, not a list. Brands that feed LLMs the right signals become the answer, which is why smaller players can now compete in spaces once dominated by marketplace giants.
Marketers often misunderstand the shift as a total departure from SEO, but the fundamentals still matter: crawlability, clarity, and content quality. What changes is the interface and the intent. LLMs parse scenarios, not strings, so your site must express the contexts your product fits—gym, study, calls, travel—through explicit, machine-readable structure. Schema and clean taxonomies help a crawler “see” your offering, and frequent updates signal relevance. Social and community content also flow into LLMs, so thoughtful posts on LinkedIn, Reddit, and owned newsletters now influence AI answers. Treat social as a distribution layer for authority, and ensure that claims made there are backed by pages your site can serve to crawlers.
A practical path forward starts with visibility analysis: where does your brand appear across ChatGPT, Gemini, Claude, Perplexity, and Google? From there, create a single source of structured truth. Introduce an llm.txt to guide crawlers, expand product and article schema, and map content to micro-intents uncovered by user queries. Query fan-outs—expanding a winning term into dozens of related conversations—reveal the long-tail prompts where you can surface quickly. Align your publishing calendar to those conversations and keep pages light, clear, and updated with verifiable details. Every change should be annotated and tracked for its impact on impressions, citations, clicks, and conversions so you can double down on what moves the needle.
Chris Donnely
Personal brand has become an accelerant for reach, validation, and feedback loops. By sharing transparent progress, frameworks, and educational content daily, founders can bootstrap distribution and recruit real users into product shaping. This isn’t about vanity; it’s about trust and access. The compounding effect is real: consistent posting drives more conversations, more data, and better insights to feed your recommendation engine. Even large brands can apply this mindset by mining on-site search data and customer queries to inform content strategy, then piping that back into prompts and structured updates. The result is an agile, human-led, AI-amplified marketing system that ships faster and learns from real signals.
AI won’t replace marketers who embrace it; it will amplify them. Use models for research, analysis, and first drafts, then add the human layer of voice, judgment, and taste. The worst AI you’ll see is today’s, which means compounding skill with these tools pays off every week. The urgent risk isn’t hallucination; it’s invisibility. If you don’t appear in the five personalized options an LLM proposes, you effectively don’t exist for that moment of demand. The path to presence is clear: structure your data, express your contexts, publish with integrity, and measure what matters. Those who move now can punch far above their weight while the window is still open.
