How AI search engines recommend businesses (and why most are invisible)
According to Semrush, AI search traffic grew 527% year-over-year in 2025. Meanwhile, 73% of businesses don't appear in a single AI-generated answer. That's not a gap. That's a disappearing act.
If you've spent years building your Google rankings, here's what you need to know: ChatGPT, Perplexity, Claude, Gemini, and Copilot don't care about your position on page one. They pick businesses using a completely different set of signals. And if you're not built for those signals, you're invisible to the fastest-growing search channel in history.
Let me walk you through exactly how each AI engine decides which businesses to recommend, what signals actually matter, and what you can do about it today.
The old playbook doesn't work here
Google ranks pages. AI engines recommend entities.
That distinction changes everything. When someone asks Google "best Italian restaurant downtown," they get a list of 10 blue links ranked by backlinks, domain authority, and keyword relevance. When someone asks ChatGPT the same question, they get a direct answer. One or two restaurants, named specifically, with reasons why.
According to Ahrefs, Google AI Overviews alone reduce organic clicks by 58%. And according to Capgemini, 58% of consumers now use AI tools instead of Google for product recommendations. Your customers are already searching differently. The question is whether AI engines are recommending you when they do.
How each AI engine picks its recommendations
Not all AI engines work the same way. Here's what's happening under the hood.
ChatGPT (OpenAI)
ChatGPT pulls from its training data, web browsing (via Bing), and increasingly from real-time search through its Browse and Plugins features. For business recommendations, it favors:
- Businesses mentioned frequently across authoritative sources (news, review sites, directories)
- Clear, factual content on your website (who you are, what you do, where you're located)
- Structured data (JSON-LD) that makes your business machine-readable
- Strong presence on directories AI engines actively crawl (Yelp, TripAdvisor, G2, Capterra)
Perplexity AI
Perplexity is citation-heavy. Every answer includes numbered source links. According to Authoritas, 79% of AI-cited sources come from just 10 domains. That means Perplexity overwhelmingly pulls from established, high-authority publishers and directories. To get cited by Perplexity, you need:
- Presence on the domains Perplexity already trusts (Wikipedia, major directories, news outlets)
- Content that reads like a reference source, not a sales page
- Entity-level clarity: Perplexity needs to understand what your business IS before it can recommend it
Google Gemini + AI Overviews
Google's AI features pull directly from its own search index. According to BrightEdge, AI-generated answers now appear in 47% of Google searches. Gemini favors:
- Strong Google Business Profile with complete, accurate information
- High E-E-A-T signals (experience, expertise, authoritativeness, trustworthiness)
- Content that directly answers questions in clear, concise language
- Schema markup that helps Google's systems parse your pages
Claude (Anthropic)
Claude draws from its training data and, when available, web retrieval. It tends to recommend businesses that have clear, well-structured informational content. Claude responds well to:
- Factual density (specific numbers, named credentials, verifiable claims)
- Content organized with clear headings and logical structure
- About pages that establish real expertise and experience
Microsoft Copilot
Copilot is built on Bing's search index combined with OpenAI models. This means:
- Bing SEO matters here (Bing Places, Bing Webmaster Tools)
- Businesses with strong Bing-indexed directory presence get favored
- Structured data and clean HTML help Copilot parse your content accurately
The 6 signals that actually matter
Across all these engines, six signals consistently determine whether your business gets recommended or ignored.
1. Entity clarity
AI engines need to understand what you are before they can recommend you. That means your website needs to clearly state your business name, category, location, and core offerings in plain language. Not clever marketing copy. Not vague taglines. Clear, factual statements.
2. Structured data (JSON-LD)
Schema markup is the language AI engines speak. LocalBusiness, Organization, Product, FAQ, and Review schemas make your content machine-readable. Without them, AI engines have to guess what your page is about. They usually guess wrong.
3. E-E-A-T signals
Experience, Expertise, Authoritativeness, and Trustworthiness. These aren't just Google concepts anymore. AI engines across the board weight content more heavily when it comes from identifiable experts with real credentials. Author bios, credentials, case studies, and "About" pages all feed E-E-A-T signals.
4. Factual density
AI engines prefer content packed with specific, verifiable facts over vague marketing language. "We serve 2,400 customers across 18 countries" beats "We serve customers worldwide" every time. Numbers, dates, named clients, specific outcomes. These are the signals that get you cited.
5. Directory presence
AI engines don't just crawl your website. They crawl directories, review sites, and aggregators. Your presence (or absence) on Yelp, TripAdvisor, G2, Capterra, Apple Maps, and industry-specific directories directly affects whether AI engines can verify and recommend you.
6. AI crawler access
Here's one most businesses miss entirely. Many websites block AI crawlers without knowing it. If your robots.txt blocks GPTBot, ClaudeBot, or PerplexityBot, those engines literally cannot index your content. You're invisible by accident.
Traditional SEO vs. GEO: what's different
| Signal | Traditional SEO | GEO (Generative Engine Optimization) |
|---|---|---|
| Goal | Rank on page 1 | Get recommended in AI answers |
| Unit of value | Pages and keywords | Entities and facts |
| Backlinks | Critical ranking factor | Less important. Citations and mentions matter more |
| Content style | Keyword-focused, often lengthy | Factual, concise, directly answering questions |
| Structured data | Nice to have | Essential. AI engines rely on it |
| Directories | Useful for local SEO | Critical. AI engines cite them heavily |
| Technical | Page speed, mobile-first | AI crawler access, robots.txt configuration |
| Measurement | Rankings, organic traffic | AI mention rate, citation presence, recommendation frequency |
| Who wins | Best-performing page | Most trusted, clearest entity |
The bottom line: SEO improves pages for ranking algorithms. GEO improves your entire digital presence for AI recommendation engines. You need both.
Why 73% of businesses are invisible
I tested over 100 businesses across all 6 major AI engines. The pattern was consistent. Businesses fail AI visibility for predictable reasons:
- No structured data. Their website has zero JSON-LD markup. AI engines can't parse what the business does.
- Blocked AI crawlers. Their robots.txt blocks GPTBot, ClaudeBot, or other AI bots. They don't even know it.
- Thin directory presence. They're on Google Business Profile but missing from the 10-15 directories AI engines actually cite.
- Vague content. Their website says "innovative solutions for modern businesses" instead of stating clearly what they do, for whom, and where.
- No E-E-A-T signals. No author bios. No credentials. No case studies. AI engines can't verify expertise, so they don't recommend it.
The fix isn't a mystery. It's a checklist. But most businesses don't know the checklist exists.
What to do about it
Start with these 5 steps:
Check your AI crawler access. Review your robots.txt for blocks on GPTBot, ClaudeBot, Bytespider, PerplexityBot, and other AI crawlers. If they're blocked, unblock them.
Add JSON-LD schema. At minimum, add Organization or LocalBusiness schema to your homepage. Add FAQ schema to your most-visited pages. Add Product schema to your product pages.
Audit your directory presence. Make sure you're listed, accurate, and complete on the directories AI engines cite most: Yelp, TripAdvisor, G2, Capterra, Apple Maps, BBB, and your industry-specific directories.
Rewrite for factual density. Replace vague marketing language with specific, verifiable statements. Add numbers. Name clients (with permission). State credentials.
Run a free AI visibility analysis. You can check how all 6 AI engines see your business right now at ChatReady's free analysis. It takes 60 seconds and shows you exactly where you're visible, where you're invisible, and what to fix first.
The window is closing
According to Semrush, AI search traffic grew 527% in a single year. According to Capgemini, 58% of consumers already prefer AI over Google for product research. This shift is accelerating.
The businesses improving their AI visibility now will own the recommendations. The rest will wonder why their traffic is declining despite solid Google rankings.
Which side do you want to be on?
Run your free AI visibility analysis and find out where you stand.