# How AI Search (ChatGPT, Perplexity, Claude) Decides Which Businesses to Recommend > AI search tools like ChatGPT and Perplexity don't use a list of links. They cite specific businesses based on structured data, entity clarity, and content quality. Learn what drives these recommendations. --- When someone asks ChatGPT "which web designer in Philadelphia should I hire" or asks Perplexity "what's the best local SEO service near me," something specific happens. The AI doesn't show ten blue links. It synthesizes an answer — and it cites specific businesses. Here's how it decides which ones. ## Retrieval vs. ranking: the fundamental difference Google ranks pages by analyzing hundreds of signals — backlinks, content relevance, page speed, user engagement — and returns a sorted list. AI answer engines like ChatGPT, Perplexity, and Claude do something different: they retrieve relevant content and synthesize it into a coherent answer. They're optimizing for what's most useful to say, not what's most authoritative to link to. This is why a business can rank on page one of Google but never appear in an AI recommendation — and why a smaller business with better-structured content can get cited by ChatGPT despite modest Google rankings. The game is different. ## Signal 1: Structured data (JSON-LD) JSON-LD is machine-readable code embedded in your website's HTML that explicitly describes what your business is. A proper LocalBusiness schema tells AI crawlers: this business is named X, it offers these services, it's located at this address, it serves this geographic area, these are its hours and contact details. Without this, AI systems have to infer your business identity from your prose — and they often get it wrong or skip you because the inference cost is too high. Most small business websites have no structured data at all. A Squarespace or Wix template might add a bare-minimum Organization schema, but it's rarely populated correctly. The competitive bar here is low — which means implementing it well is a significant advantage. ## Signal 2: FAQ schema When someone asks an AI tool a question, the AI is looking for content structured as questions and answers. FAQ schema is markup that explicitly tags your content as Q&A pairs, making it trivial for AI systems to extract and use. If your website has a FAQ section that isn't marked up with schema, AI tools can't reliably identify it as authoritative Q&A content versus regular paragraph text. FAQ schema is also one of the signals Google uses to generate FAQ rich results in traditional search — the expandable question-answer dropdowns you sometimes see in search results. It's one of the few optimizations that simultaneously benefits Google rankings and AI citation likelihood. ## Signal 3: llms.txt llms.txt is an emerging standard — a plain-text file served at yoursite.com/llms.txt that provides AI crawlers with a structured summary of your business: who you are, what you do, your services and pricing, common questions and answers, and how to contact you. Think of it as the resume you hand an AI system so it doesn't have to reverse-engineer your business from your website's marketing copy. Perplexity, ChatGPT's web search component, Claude.ai, and other AI systems that actively crawl the web will find and use this file. Most websites don't have one. Every site I build ships with a populated llms.txt. For existing sites, adding it takes about an hour and is one of the highest-leverage AEO actions you can take right now. ## Signal 4: Entity clarity AI systems build a mental model of your business as an entity — a named thing with consistent attributes. The clearer and more consistent your business identity is across your website, Google Business Profile, and the broader web, the more confidently an AI can reference you. Inconsistencies — slightly different business names, different phone numbers, different service descriptions — create ambiguity that makes AI systems less likely to cite you with confidence. This is why local SEO fundamentals — consistent NAP (name, address, phone) data, a well-maintained Google Business Profile, accurate directory listings — matter for AI search, not just Google Maps. Entity clarity is the shared foundation. ## Signal 5: Content extractability AI systems need to be able to read your content. This sounds obvious, but many websites fail on it: text embedded in images, content loaded via JavaScript that crawlers can't access, animated sliders with no text fallback, or copy so abstract ("we deliver innovative solutions for forward-thinking businesses") that there's nothing concrete for an AI to extract. The content that gets cited by AI tools is specific: this business is located at X address, it serves Y area, it offers Z services at these price points, here's what clients say. Specific beats vague every time. Direct beats clever. The marketing copy written to impress humans often reads as empty noise to an AI retrieval system. ## What this means for your business right now The competitive landscape for AI citations is still wide open. Most small businesses haven't thought about any of this — which means the ones that act first build a meaningful lead. I run Complete SEO/AEO audits ($595) that explicitly evaluate all five of these signals and tell you what to fix. I also build new sites that implement all of these from day one, so you launch discoverable rather than invisible. The window to be an early mover on AI search optimization for your local market is still open. It won't be in 18 months. The businesses that get structured now will be the ones AI tools recommend when your future customers start asking. --- **Contact Information:** - Email: pablo@heymoai.com - Address: Philadelphia, PA, US