How Does ChatGPT Decide Who to Recommend?
Key Takeaways
- The Mechanism: Unlike Google, ChatGPT asks qualifying questions to match intent to expertise before recommending anyone.
- The Data Source: Over 70% of local business results in ChatGPT come from Foursquare data, not Google Maps.
- The Strategy: Businesses must position themselves to answer the “hidden” questions AI asks (e.g., “Emergency or Renovation?” not just “Plumber”).
- The Gap: AI traffic converts at 14.2% (vs 2.8% on Google) because the user is already “interviewed” and qualified by the AI.
The Short Answer
ChatGPT doesn’t rank websites; it has conversations. Before recommending a business, it acts like a consultant: it asks clarifying questions to understand the user’s specific situation, then matches that intent to the businesses with the clearest, most specific positioning. It doesn’t look for the “best” business; it looks for the most confident match.
Why This Question Matters Now
A potential customer opens ChatGPT and types: “I need a massage in Austin.”
Here’s what Google does: Shows 10 links to massage parlors and lets you figure it out.
Here’s what ChatGPT does: It interviews you.
If you’ve shared health goals with ChatGPT before, it leverages that context:
Only after understanding your actual intent does it recommend specific businesses.
This is fundamentally different from how search has worked for 25 years. And it changes everything about how you need to position your company.
ChatGPT Doesn’t Just Answer—It Interviews
This is the insight most people miss: AI rarely answers cold requests directly.
When a user asks for a recommendation, ChatGPT behaves more like a knowledgeable friend than a search engine. It hunts for specificity:
- For a Plumber: “Emergency repair or bathroom renovation?”
- For an Accountant: “Personal tax, business audit, or bookkeeping?”
- For a Lawyer: “Family law, estate planning, or litigation?”
- For a Massage Therapist: “Relaxation, injury recovery, or prenatal care?”
Why This Changes Everything
On Google, a massage business tries to rank for “massage Austin”—a broad keyword.
On ChatGPT, that same query triggers a conversation. When the user says “I’m recovering from a marathon,” ChatGPT filters out the “Generic Day Spas” and looks for businesses positioned as “Sports Recovery Specialists.”
The business that says “we do everything for everyone” is now losing to the one that says “we specialize in X for Y.”
AI rewards specificity because specificity allows it to make a confident match.
AI Knows More Context Than You Think
Here’s another layer most people don’t consider: ChatGPT often has context about the person asking.
If someone has previously discussed:
- Training for a marathon
- Dealing with chronic back pain
- Being pregnant
- Running a small business
- Moving to a new city
ChatGPT factors that into its recommendations. It’s not starting from zero—it’s building on what it already knows about the user’s situation.
This means when someone asks “I need a massage,” ChatGPT might already know they’re an athlete recovering from an injury, or a pregnant woman in her third trimester, or a stressed-out founder who mentioned burnout last week.
The businesses that get recommended are the ones whose positioning matches the context AI already has.
The “Cheat Code”: Answering the Invisible Questions
Here is how smart businesses are winning right now: Anticipate the interview.
Think about the questions ChatGPT has to ask a user before recommending you. Then, ensure your digital presence answers those questions explicitly.
The “Plumber” Test
If ChatGPT asks the user: “Is this residential or commercial?” and “What specific suburb?”
Your website needs to say:
- “Specializing in Residential renovations…”
- “Serving Manningham and Whitehorse councils…”
When ChatGPT sees that data, it doesn’t just see a keyword; it sees the answer to its own question. It can recommend you with zero hallucination risk.
The businesses that win in AI search are the ones that answer the qualifying questions before they’re asked.
The Foursquare Factor (Yes, Really)
Here’s the data point that surprises everyone:
Over 70% of local business results shown in ChatGPT queries come from Foursquare data. (GMBAPI, August 2025)
Yes, Foursquare. Even though the consumer app faded, their data layer powers the “Places” API for OpenAI.
- When you ask “cafés near me,” ChatGPT pings Foursquare.
- Ratings, photos, and price tiers? Foursquare.
- Map visualizations? Mapbox (often powered by Foursquare data).
The Trap: If your business has a perfect Google Business Profile but an unclaimed or outdated Foursquare listing, you might be invisible to ChatGPT for local queries.
The Six Factors ChatGPT Uses to Decide
We have analyzed the architecture of how LLMs synthesize recommendations. Here is the matrix of what matters:
| Factor | Weight | What AI Looks For |
|---|---|---|
| 1. Clear Positioning | Critical | Does the business define who they serve? (e.g., “Renovations” vs “Plumbing”) |
| 2. Entity Identity | Critical | Is there Schema Markup defining the Founder, Credentials, and Service Areas? |
| 3. Answer Formatting | High | Are there FAQs structured as direct answers to specific questions? |
| 4. Review Signals | High | Do reviews mention specific outcomes? (“Fixed my slab leak” vs “Great job”) |
| 5. 3rd Party Validation | High | Is the business cited on authoritative industry sites? |
| 6. Directory Data | Med-High | Is the Foursquare/Mapbox data accurate and consistent? |
Visualizing the Difference
Here is what the difference looks like when an AI processes your business:
❌ The “Undercooked” Business
Positioning: “Full-service massage and wellness center.”
The Scene: User tells ChatGPT: “I need sports recovery after a marathon.”
The Result: ChatGPT skips this business. It cannot be sure if they handle sports injuries. It is a “low confidence” match.
✅ The “AI-Ready” Business
Positioning: “Sports massage and deep tissue therapy for endurance athletes.”
The Scene: User tells ChatGPT: “I need sports recovery after a marathon.”
The Result: ChatGPT sees a 100% intent match. It recommends immediately: “For post-marathon recovery, I recommend [Name]—they specialize in endurance athletes.”
How to Check If ChatGPT Can Match You
Here’s a simple test you can run right now:
- Open ChatGPT and ask for a recommendation in your category
- Watch what qualifying questions it asks
- Ask yourself: Does my website clearly answer those questions?
If ChatGPT asks “What type of massage—relaxation or therapeutic?” and your website says “we offer relaxation AND therapeutic massage,” you haven’t helped AI make a confident match.
But if your website says “We specialize in therapeutic deep tissue massage for chronic pain and injury recovery,” ChatGPT knows exactly when to recommend you.
Then try: “What do you know about [Your Business Name]?”
If AI doesn’t recognize you as a specific entity with clear expertise, you have an AI visibility problem—regardless of how well you rank on Google.
Case Study: Harmony Group
We proved this with Harmony Group—a co-living property investment firm less than 12 months old that went from invisible to #1 AI recommendation in Australia in just six weeks.
Not because they had more content than competitors. Because when ChatGPT asked itself “who specializes in 1B certified co-living?” they were the only clear answer. Their positioning matched the qualifying question perfectly.
The Bottom Line
The businesses currently winning on ChatGPT aren’t necessarily the “best”—they are the easiest to categorize.
They are winning because when the AI asks itself “Is this the right match for this specific user?”, their content provides a clear “Yes.”
The era of ranking for keywords is ending.
The era of answering the qualifying question has begun.
Frequently Asked Questions
Why does ChatGPT ask questions instead of just recommending?
Because intent matters. “I need a lawyer” is too vague. ChatGPT asks clarifying questions to ensure it doesn’t recommend a divorce lawyer to someone buying a house. It mimics how a knowledgeable human consultant would help you find the right match.
Does ChatGPT remember what I’ve told it before?
Yes. If you previously mentioned you are a business owner, training for a marathon, or dealing with back pain, ChatGPT retains that context through its memory features to personalize future recommendations without asking you again.
Why does Foursquare matter if no one uses it anymore?
The consumer app declined, but the database is the industry standard for enterprise API data. OpenAI uses the Foursquare Places API to retrieve real-world business data. Over 70% of local business results in ChatGPT come from Foursquare. If you ignore it, you ignore the data source feeding the AI.
How do I know what qualifying questions ChatGPT will ask?
Test it yourself. Open ChatGPT, use a temporary chat, and ask for a recommendation in your industry. Note the specific follow-up questions it asks you. Those questions are your roadmap for what content to build and how to position your business.
What is the ROI of AI search visibility?
AI search traffic converts at 14.2% compared to 2.8% for traditional Google traffic—a 5x improvement. You get fewer visitors, but they are pre-qualified, pre-educated through the AI conversation, and ready to buy.
How long does it take to start appearing in AI recommendations?
With the right infrastructure in place, we typically see first AI citations within 60-90 days. Harmony Group hit #1 in their category in Australia in just six weeks. The timeline depends on your starting point and the competitiveness of your niche.
Probably Genius helps high-trust professional service firms become the answer AI gives.
If you want to understand exactly how AI sees your business—and what qualifying questions you need to answer—
