How ChatGPT decides which businesses to recommend
There's no ranking page to inspect and no keyword to chase. But the recommendation isn't random — it's assembled from evidence. Here's the evidence.
AI assistants recommend businesses by combining three layers of evidence: what they learned in training (your long-term reputation), what they find in live search at answer time (your current pages and citations), and structured sources like maps, directories and reviews. You influence all three the same way — a consistent, well-evidenced public record.
The three layers behind every recommendation
Models learn about businesses from years of web text: articles, directories, forums, reviews. If your brand has a thin or inconsistent public record, the model simply knows less about you — and recommends what it knows.
Most assistants now run a web search before answering and ground their recommendation in what comes back. Your current rankings, pages and citations feed this layer — this is where SEO and AEO overlap most.
For local and service queries, assistants lean on structured data: business profiles, ratings, hours, categories. Wrong category or stale profile here quietly disqualifies you from whole classes of questions.
What the model is actually weighing
When an assistant assembles a shortlist, the businesses that surface share a pattern: they're described the same way everywhere (name, category, location, specialty), they're mentioned by sources the model already trusts (industry publications, associations, high-authority directories), they have review evidence that is recent and substantial, and their own sites answer questions directly in a form a machine can lift. None of these is a trick. They're all verifiable facts about your public record.
Why you might be invisible — or worse, wrong
Absence is usually an evidence problem: too few independent mentions for the model to be confident about you. Misdescription is worse — it means the evidence conflicts. An old directory listing says you're a "web design agency", your LinkedIn says "digital marketing", your site says "AI visibility" — the model averages the confusion and either hedges or skips you. The fix is the least glamorous work in marketing: making the public record agree with itself.
How to influence the answer
Fix the public record first
Same name, category, offer and location on your site, Google Business Profile, LinkedIn, directories and associations. Boring, decisive.
Mark up the facts
Organization, Service, FAQPage and Review schema turn claims into machine-readable facts assistants can quote confidently.
Earn mentions where models read
Industry publications, professional associations, credible directories, podcasts with transcripts — sources with authority in your niche.
Publish answer-shaped pages
One real customer question per page, answered directly in the first paragraph, with evidence below. These get lifted verbatim.
Monitor the answers themselves
Ask the assistants your customers' questions monthly. Track who gets named, how you're described, and what changed — then adjust.
Want to see where you stand right now? Run the 21-point AI visibility checklist, or read the complete AEO guide.
Common questions
Does ChatGPT recommend businesses?
Yes. When asked for providers, products or local services, ChatGPT and similar assistants name specific businesses, drawing on training data, live web search and structured sources like maps and review platforms.
Can you pay to be recommended by AI?
No. There is no paid placement inside organic AI answers today. Recommendations are earned through consistent business data, trusted third-party mentions, reviews and quotable content.
Why does AI describe my business incorrectly?
Usually because the public record is inconsistent — conflicting categories, outdated directories, thin structured data. Fixing entity consistency is the fastest way to correct AI's understanding.
Playbooks by industry
The signals are universal; the questions are not. Each playbook covers what customers ask in one industry and how the answer gets decided.
Keep reading:how Google AI Overviews work, AEO vs SEO vs GEO, and the AI search glossary.