Why ChatGPT Recommends Some Restaurants and Not Yours

It is Friday night. A couple visiting your city opens ChatGPT and types "best Italian restaurant in [your neighborhood] with outdoor seating." ChatGPT names three places. You have been serving handmade pasta for 12 years and you are not on the list.

Restaurants are the single most common local business query on AI. More people ask ChatGPT and Gemini for restaurant recommendations than any other business category. And the gap between restaurants that show up and restaurants that do not is widening fast.

The restaurants winning AI recommendations are not necessarily the best restaurants. They are the restaurants with the best data.

How People Search for Restaurants on AI

Restaurant queries on AI are wildly different from Google searches. Nobody types "restaurant 90254" into ChatGPT. They describe what they want like they are talking to a friend who knows the area.

Occasion-based queries

• "Romantic restaurant for an anniversary dinner in [city], not too loud"
• "Best place for a birthday dinner for 12 people near [area]"
• "Casual brunch spot near [neighborhood] that does not need reservations"

Cuisine-specific queries

• "Authentic Mexican food in [city], not a chain"
• "Best sushi near me that is not all-you-can-eat"
• "Where to get the best pho in [city]?"

Dietary and preference queries

• "Vegan-friendly restaurant in [area] that is not just salads"
• "Gluten-free options near me with good reviews"
• "Restaurant with a good kids menu near [neighborhood]"

Discovery queries

• "Hidden gem restaurants in [city] that tourists do not know about"
• "New restaurants that opened this year in [area]"
• "Best restaurant in [city] under $30 per person"

These queries are specific. They mention ambiance, cuisine type, dietary needs, party size, price range, and occasion. If your restaurant's AI profile does not include these details, you cannot match these queries. Full stop.

Why Foursquare Matters More for Restaurants Than Any Other Business

Foursquare feeds ChatGPT's local recommendations. For most businesses, Foursquare is one data source among several. For restaurants, it is the data source.

Foursquare has the most granular restaurant category system in existence. Over 400 food-related subcategories. Not just "Restaurant" but "Neapolitan Pizza Restaurant," "Ramen Restaurant," "Farm-to-Table Restaurant," "Tapas Restaurant," "BBQ Joint."

When someone asks ChatGPT "best Thai restaurant near me," the AI matches that query to Foursquare's "Thai Restaurant" category. If your Thai restaurant is categorized as just "Restaurant" on Foursquare, you will not match that query. It is that simple.

Here is what your Foursquare restaurant listing needs:

1. Specific cuisine category as primary. "Italian Restaurant," "Sushi Restaurant," "Mexican Restaurant." Not "Restaurant."
2. Secondary categories for atmosphere. "Outdoor Dining," "Wine Bar," "Family Restaurant" as applicable.
3. Price tier set correctly. Foursquare uses a 1-4 price scale. This matches price-sensitive queries.
4. Menu highlights in your description. Signature dishes, dietary options, notable ingredients.
5. Hours fully completed. Brunch hours, late-night hours, happy hour. All of them.
6. Photos of actual dishes. Not stock photos. Real food from your kitchen.

The Menu Data Gap

AI models want to recommend specific dishes. When someone asks "where can I get a good carbonara near me," the AI is looking for restaurants that have carbonara in their structured data.

Most restaurants put their menu on their website as a PDF. AI cannot read PDFs. Some restaurants use third-party menu platforms that block web crawlers. AI cannot access those either.

Your menu needs to be on your website as actual HTML text, not a PDF or an embedded image. Every dish should be searchable text with a description and price. This is the single biggest thing most restaurants get wrong for AI visibility.

If you use a platform like Square or Toast for online ordering, check whether your menu is rendered as crawlable HTML. Many of these platforms use JavaScript rendering that AI web crawlers cannot access.

Photos: Volume and Recency Beat Quality

For restaurants, photos matter more than for almost any other business category. But not for the reason you think.

AI models do not evaluate photo quality directly. They cannot tell if your food photography is good. But Google's Gemini and AI Overviews pull from Google Maps, and Google Maps uses photo volume and recency as ranking signals.

A restaurant with 500 user-uploaded food photos posted over the past 6 months will outrank a restaurant with 20 professional photos from 3 years ago. Recency signals to AI that the restaurant is active and popular.

Encourage customers to post food photos on Google. Add a small card to tables: "Love your meal? Share a photo on Google Maps." Every photo is a data point that strengthens your AI presence.

On your own Google Business Profile, post food photos weekly. New menu items, daily specials, seasonal dishes. Each post tells Google (and by extension Gemini) that your restaurant is active.

Review Keywords That Trigger Restaurant Recommendations

AI models extract specific signals from restaurant reviews. Star ratings are the baseline, but review text is where the matching happens.

• Dish mentions: "The truffle pasta was incredible" matches dish-specific queries
• Ambiance descriptors: "Quiet enough for conversation," "great patio," "perfect for date night"
• Service notes: "The server knew the wine list," "no rush, very relaxed pace"
• Dietary mentions: "Great vegan options," "they were very accommodating with allergies"
• Value indicators: "Worth every penny," "best value in the neighborhood"
• Occasion tags: "Celebrated our anniversary here," "perfect for a business lunch"

When someone asks ChatGPT "quiet restaurant for a date in [city]," the AI scans review text for "quiet," "date night," "romantic," and "intimate." If your reviews have those words, you match. If they do not, you do not.

Google Maps and Gemini: The Direct Pipeline

Your Google Business Profile feeds Gemini directly. Google's "Ask Maps" feature uses AI to answer restaurant queries with data from your GBP. This is already live and growing.

For restaurants, the GBP fields that matter most are:

• Menu link: Add your actual menu URL. Gemini can crawl and reference it.
• Attributes: "Outdoor seating," "Reservations required," "Good for groups," "Takeout available." Check every one that applies.
• Popular dishes: Google now lets you highlight popular dishes on your profile. Use this feature.
• Posts: Weekly posts about specials, new dishes, or events. Active profiles rank higher.

The Reservation and Ordering Integration

AI models are starting to integrate with reservation and ordering systems. When someone asks "book a table at [restaurant name]," ChatGPT can link to OpenTable or Resy. When someone asks "order delivery from [restaurant]," Gemini can link to your ordering system.

If you use OpenTable, Resy, Yelp Reservations, or a direct booking system, make sure it is linked from your Google Business Profile and your website. AI models use these integrations as a quality signal. A restaurant with a working reservation link is a verified, active business.

The Numbers for Restaurant Owners

An average dinner cover is $40 to $80 per person. A table of four is $160 to $320. If AI search sends you just one extra table per night that you would not have gotten otherwise, that is $4,800 to $9,600 per month in additional revenue.

For a restaurant, AI visibility is not a "nice to have." It is a revenue channel. And right now, most restaurants in your area have terrible AI data. Their Foursquare listings are unclaimed, their menus are PDFs, and their Google Business Profiles are half-filled-out.

That is your window. Set up your three-platform presence, get your menu into crawlable HTML, and encourage reviews that mention specific dishes. Or let PACO GEO handle the optimization for you.