Ask ChatGPT "who's the best roofer in Cumming, GA?" and it will give you names. Specific businesses, often with a sentence about why each one made the list. For the businesses named, that's a referral that cost nothing. For everyone else, it's a customer they never knew they lost.
So how does ChatGPT decide? It's not random, it's not paid placement, and it's not just whoever ranks first on Google. Here's what's actually happening.
AI doesn't crawl the web in the moment. It recalls and verifies.
When ChatGPT answers a local recommendation question, it draws on two things: what its underlying model learned about your market during training, and what it can verify through live web search at the moment of the question. Both paths reward the same fundamentals.
The model's training data includes review platforms, business directories, local news, community discussions, and your own website. If your business appears consistently across those sources with strong sentiment, the model "knows" you. If you exist mostly as a Google Ads budget and a thin website, you're invisible to it.
When ChatGPT searches live, it behaves more like a skeptical researcher than a search engine. It pulls several sources, cross-references them, and synthesizes an answer. Businesses that appear in multiple trustworthy places win. Businesses that appear in one place, or appear with conflicting information, get filtered out.
The five signals that matter most
Across the audits we run for Forsyth County businesses, the same factors separate the recommended from the invisible:
Review depth and recency. Not just your star rating: volume, recency, and what reviewers actually say. A 4.8 with 200 reviews mentioning specific services beats a 5.0 with 12 reviews from 2022. AI reads review text, so "they replaced our roof in Cumming after the hailstorm" is a signal a star count can't deliver.
Citation consistency. Your name, address, and phone number across Google Business Profile, Yelp, Bing Places, Apple Maps, the BBB, and industry directories. When these match, AI gains confidence you're a real, stable business. When they conflict, confidence drops, and so does your recommendation rate.
Third-party mentions. Local news coverage, chamber of commerce listings, community sites, sponsorships. AI treats these as independent validation, much like a journalist trusts a source that other sources corroborate.
Website clarity. Can a machine reading your site instantly tell what you do, where you serve, and why you're credible? Clear service pages, location language, FAQs, and structured data all help. Vague taglines and image-only content don't.
Entity coherence. All of the above telling the same story. One business, one identity, consistent everywhere.
What this means if you're not being recommended
The encouraging part: these are all fixable. Unlike traditional SEO, where outranking an entrenched competitor can take years, AI visibility responds to trust signals that most local businesses have simply never managed deliberately. In market after market, we find the AI-recommended business isn't the biggest one. It's the most verifiable one. For a deeper breakdown of GEO, AEO, and how they differ from traditional SEO, see our complete guide to GEO, AEO, and local AI visibility.
The first step is knowing where you stand. Run the question yourself: ask ChatGPT, Gemini, and Perplexity who they'd recommend in your category in Forsyth County, and see if you're in the answer. If you're not, our Forsyth County research reports show exactly how these recommendations break down by industry, and a free AI Visibility Audit will show you the specific gaps for your business.
The businesses winning AI recommendations right now aren't winning because they figured out a trick. They're winning because almost nobody else in their market is paying attention yet. That window won't stay open.