How AI Assistants Decide Which Restaurants and Bars to Recommend

What Is GEO — and Why Should Restaurant and Bar Owners Care?

GEO stands for Generative Engine Optimisation. It is how you get your restaurant or bar recommended by AI assistants like ChatGPT, Perplexity, and Google AI Overviews.

Think about how you used to search. You typed "best Italian restaurant Manchester" into Google and got ten blue links. You clicked three, compared menus, checked reviews, and made a decision. That entire process is collapsing into a single step.

Now a customer types the same question into ChatGPT or Perplexity and gets one direct answer: a name, a reason, and a link. No list of ten. No second page. If AI does not mention your venue, you do not exist in that conversation.

This is not a future prediction. It is happening right now. 40% of Gen Z already use AI chatbots for restaurant and bar recommendations before they open Google (Bain & Company / Malou, 2025). Across all age groups, 20% of consumers now use AI tools for restaurant discovery (Bain & Company, 2025). Gartner predicts that 25% of organic search traffic will shift to AI chatbots by the end of 2026.

Traditional SEO focused on ranking your website in a list. GEO focuses on making your venue the answer. The techniques overlap — structured data, reviews, fresh content — but the goal is different. You are not competing for position seven. You are competing for the one recommendation an AI gives.

For independent restaurant and bar owners, this is both a threat and an opportunity. The threat: chains with marketing teams are already optimising for AI. The opportunity: most independents have not started, which means the first movers in your area will dominate AI recommendations before competitors even notice.

The question is no longer whether your customers use AI to find restaurants. It is whether AI can find yours.

How Do AI Assistants Choose Which Venues to Recommend?

Each AI platform pulls from different sources, but they all share one thing in common: they need structured, up-to-date information about your restaurant or bar. If that information does not exist, neither do you.

ChatGPT and the Bing Connection

ChatGPT does not crawl the web independently. It relies on Bing's search index for real-time information. A study by Seer Interactive in 2024 found that 87% of ChatGPT's citations match Bing's top 10 organic results. If your restaurant or bar is invisible on Bing, it is invisible on ChatGPT. Most UK venue owners have never claimed their Bing Places listing. That is a problem we will fix in Step 2.

Perplexity and Niche Authority

Perplexity works differently. It does not just mirror mainstream search results. 24% of Perplexity's citations come from industry-specific sources like TripAdvisor, Time Out, and local food blogs — not generic search pages (Seer Interactive, 2024). This means your TripAdvisor profile and any mentions on niche food sites carry outsized weight on Perplexity.

Google AI Overviews

Google's AI Overviews now appear in 15-25% of searches and that number is climbing. Their system uses a technique called "query fan-out" — it does not just answer your exact question. It searches related topics, pulls context from multiple sources, and synthesises a single response. For a query like "romantic restaurant Manchester," Google's AI might simultaneously check your Google Business Profile, review sentiment, menu data, and local blog mentions.

Ask Maps: The Game Changer

Ask Maps is Google's AI layer inside Google Maps. It reads your Google Business Profile data — categories, attributes, photos, reviews, and owner responses — to generate conversational recommendations. If your profile is missing opening hours, menu links, or photo descriptions, Ask Maps has nothing to work with and will recommend the venue next door instead.

Freshness Matters More Than Ever

AI assistants heavily penalise stale information. 76.4% of ChatGPT's top-cited pages were updated within the last 30 days (Seer Interactive, 2024). A restaurant or bar with a Google profile last updated six months ago is competing against venues that update weekly.

What Our UK Audit Found

We analysed 2,743 restaurant and bar listings across Manchester, London, Leeds, and Birmingham. The results were stark: 62% have incomplete Google Business Profiles. 98% are missing structured menu data that AI assistants can read. And 8% are completely invisible to AI — no structured data anywhere on the web. These are not obscure venues. Many have strong local reputations but zero digital presence where AI is looking.

Step 1: Complete Your Google Business Profile

Your Google Business Profile is the single most important piece of digital infrastructure for your restaurant or bar. It is the first place Ask Maps looks. It feeds Google AI Overviews. And it influences what Perplexity and ChatGPT surface through aggregated review data.

Our audit found that 62% of UK venues have incomplete profiles. The most common gaps are missing categories, empty attribute fields, and outdated descriptions. Each gap is a missed signal to AI.

Here are the five fields AI assistants actually read:

1. Categories. Your primary category ("Italian Restaurant" or "Cocktail Bar") tells AI what you are. Secondary categories ("Pizza Restaurant," "Wine Bar") tell it what else you offer. Most venues only set one category. Set all that apply.

2. Attributes. These are the yes/no fields that answer specific questions: outdoor seating, wheelchair access, free Wi-Fi, accepts reservations. For bars specifically, attributes like "beer garden," "live music," "real ale," and "pub quiz" are the exact terms customers use when asking AI for recommendations. Venues with 10+ attributes filled are 2.3x more likely to appear in AI recommendations than venues with fewer than 5 (booteek audit, 2026).

3. Description. You have 750 characters. Use them. Include your cuisine type, neighbourhood, price range, and what makes you different. Write it as a sentence a human would say, because that is exactly how AI will quote it.

4. Photos. Venues with more than 25 photos receive 35% more direction requests than those with fewer than 10 (Google Business Profile Insights data). AI uses photo metadata and alt text to understand your venue. Label your photos with descriptive file names before uploading.

5. Q&A. The questions and answers section on your Google profile is directly indexed by AI. Seed it yourself with common questions: "Do you take walk-ins?" "Is there parking nearby?" "Do you have a children's menu?" The answers you provide become the answers AI gives.

For bar owners: do not assume your profile is complete because you set a category. Check your attributes for bar-specific fields — happy hour, cocktail menu, draught beer selection, live sport, late night opening. These are the signals that separate you from a generic "bar" listing.

Step 2: Claim Your Bing Places Listing

This is the single highest-impact action most restaurant and bar owners have never taken. 87% of ChatGPT's citations come from Bing's index (Seer Interactive, 2024). If you are not on Bing Places, you are invisible to ChatGPT.

Bing Places for Business works almost identically to Google Business Profile. You claim your listing, verify ownership, and fill in the same fields: categories, hours, photos, description, and menu links. The entire process takes about 15 minutes.

Most UK venue owners do not even know Bing Places exists. In our audit, fewer than 12% of the 2,743 venues had claimed their Bing Places listing. That means 88% of restaurants and bars in our sample are invisible to the AI assistant with the largest user base in the world.

Bing also feeds Microsoft Copilot, which is embedded into Windows, Edge, and Office 365. As Microsoft rolls AI into more products, your Bing Places listing becomes the foundation for visibility across the entire Microsoft ecosystem.

Action: Go to bingplaces.com, search for your venue, and claim it. If it does not appear, create a new listing. Bing can import directly from your Google Business Profile, so you do not need to enter everything twice.

For a step-by-step walkthrough, see our Bing Places guide.

Step 3: Replace PDF Menus with Structured Data

98% of UK venues in our audit are missing structured menu data. Most have a PDF menu uploaded to their website or Google profile. The problem: AI cannot read PDFs. Not reliably. Not consistently. A PDF menu is a black box to ChatGPT, Perplexity, and Google AI Overviews.

When a customer asks an AI assistant "What vegan options does The Fox and Hound have?", the AI needs to find that information in a format it can parse. A PDF menu with a scanned image of handwritten specials is useless. A structured HTML page with dish names, descriptions, prices, dietary labels, and allergen information is exactly what AI needs.

This also gives you a compliance advantage. UK allergen law (Natasha's Law, 2021) requires food businesses to provide full ingredient and allergen information for pre-packed foods. A structured digital menu with allergen tags is not just AI-friendly — it demonstrates regulatory compliance.

For restaurants: List every dish with a name, short description, price, and dietary tags (vegan, vegetarian, gluten-free, dairy-free). Group by course.

For bars: Do the same for your drinks menu. List cocktails with ingredients and prices. List draught beers with ABV and brewery. List wines by region and glass/bottle prices. If you serve food, structure that separately.

You do not need a developer. Services like MenuFox, BentoBox, and even a well-structured WordPress page will work. The key is that the content is in HTML text — not embedded in an image or a downloadable file.

In Step 4, we will show you the schema markup that wraps this menu data in a format AI can read instantly.

Step 4: Add Schema Markup to Your Website

Schema markup is a standardised code format that tells AI assistants exactly what your venue is, where it is, and what it offers. Think of it as a machine-readable business card embedded in your website's code.

Without schema markup, AI has to guess. It scrapes text from your website, tries to figure out whether "The Albion" is a restaurant name or a street name, and often gets it wrong. With schema markup, there is no guessing. You tell AI directly: this is a restaurant, it is at this address, it serves this cuisine, these are the opening hours, and here is the menu.

A study by Schema App found that pages with structured data receive 40% more clicks from search results — and that advantage compounds with AI, because AI assistants prefer sources with unambiguous structured data.

Below are three copy-paste code blocks for the most common UK venue types. Add the relevant one to the `` section of your website's homepage. If you use WordPress, plugins like Yoast SEO or Rank Math can add schema without touching code. If you use Squarespace or Wix, paste it into the custom code injection area in your site settings.

Each block includes UK-specific fields like `addressCountry: "GB"`, local currency formatting, and cuisine types relevant to UK venues.

Step 5: Build Conversational FAQs

AI assistants answer questions. If your website already contains those answers in a question-and-answer format, AI is far more likely to quote you directly.

Think about what customers actually ask before visiting your restaurant or bar. Not marketing questions — real operational questions. "Do you have a beer garden?" "Can I book for a group of 12?" "Do you serve Sunday roast?" "Is there live music on Saturdays?" "Do you cater for coeliac dietary requirements?"

Pages with FAQ schema markup are 2x more likely to be featured in AI-generated answers (Search Engine Journal, 2025). That means your FAQ page is not just a customer service tool — it is an AI visibility asset.

Write 10-15 FAQs in natural, conversational language. Answer each question in 2-3 sentences. Avoid jargon. Write as if you are speaking to a first-time visitor.

Here are examples that work well for UK restaurants and bars:

• "Do you have a beer garden?" → "Yes, our beer garden seats 40 and is open from April to October, weather permitting. Dogs are welcome in the garden area."
• "Do you serve Sunday roast?" → "We serve a traditional Sunday roast every week from 12pm to 4pm. Choose from beef, lamb, or a vegetarian nut roast, all with Yorkshire pudding and seasonal vegetables."
• "Can I book for large groups?" → "We can accommodate groups of up to 25 in our private dining area. For groups over 12, please call ahead to arrange a set menu."

Once written, add FAQ schema markup (FAQPage type) to the page. This tells AI assistants that the content is in question-answer format and can be quoted directly.

Step 6: Respond to Reviews Strategically

Reviews are not just social proof for human customers. They are training data for AI. When ChatGPT or Perplexity recommends a restaurant or bar, it often cites review content — paraphrasing themes, quoting sentiment, and referencing star ratings.

A study by MyPlace found that venues with 3.6x more reviews than the local average are significantly more likely to appear in AI recommendations. Volume matters. But so does recency. AI assistants weight recent reviews far more heavily than reviews from two years ago.

Response rate matters as much as review count. Google's own documentation confirms that businesses that respond to reviews are considered more trustworthy. Our audit found that only 23% of UK venues respond to more than half their reviews. The ones that do have measurably higher AI visibility scores.

Here is the strategy that works:

Respond within 24 hours. AI indexes review responses. A fast response signals an active, engaged business. A review sitting unanswered for weeks signals neglect.

Reference specific details. Do not write "Thanks for visiting!" Write "Glad you enjoyed the beef short rib — our chef sources it from Cumbrian farms." Specific details give AI more structured information to work with.

Address negative reviews professionally. AI reads negative reviews too. A calm, solution-oriented response to a complaint tells AI (and customers) that you take feedback seriously. Never argue. Acknowledge, explain, and invite the customer back.

Mention what you offer. Naturally weave in keywords: "Thanks for mentioning our weekend brunch — we have just added a new cocktail pairing menu." This is not keyword stuffing. It is providing information that AI can use to match you with future queries.

For bar owners: review responses are especially valuable because bar-specific queries ("best craft beer bar Leeds") have less competition in AI than restaurant queries. A well-maintained review profile can put you at the top of a category with fewer competitors.

For more detailed guidance on review response techniques, see our review response guides.

Step 7: Expand Your Digital Footprint

AI assistants cross-reference multiple sources. The more places your venue appears with consistent, accurate information, the more confident AI becomes in recommending you. Inconsistent data — different opening hours on Google and TripAdvisor, a wrong phone number on Bing — actively damages your AI visibility.

Prioritise these platforms in order:

• Google Business Profile — The foundation. Already covered in Step 1.
• TripAdvisor — Perplexity's second most-cited source for hospitality. 24% of its citations come from niche sites like TripAdvisor (Seer Interactive, 2024). Claim your listing, respond to reviews, and keep photos current.
• Bing Places — Feeds ChatGPT and Microsoft Copilot. Covered in Step 2.
• Apple Business Connect — Apple Maps is the default on every iPhone. Apple Maps handles 30% of mobile navigation queries in the UK (Counterpoint Research, 2025). Claim your listing at businessconnect.apple.com.
• OpenTable / TheFork — If you take reservations, these platforms provide structured booking data that AI uses to confirm operational details.

The key principle: consistency across every platform. Same name, same address, same phone number, same opening hours. AI penalises conflicting information more harshly than missing information.

Aim to have your venue accurately listed on at least 4 of these 5 platforms within the next 30 days.

Step 8: Measure Your AI Visibility

You cannot improve what you do not measure. Here is a simple self-audit you can do right now, without any tools:

The 5-Minute AI Visibility Test:

• Open ChatGPT and ask: "What are the best [your cuisine] restaurants in [your neighbourhood]?" Does your venue appear?
• Ask Perplexity the same question. Note whether it cites TripAdvisor, Google, or a local blog.
• Google "[your venue name] [your city]" and check whether an AI Overview appears at the top. What information does it show?
• Ask ChatGPT: "Tell me about [your venue name]." Does it know you exist? Is the information accurate?
• Repeat for bar-specific queries: "Best cocktail bars [your city]," "Pubs with beer gardens [your area]."

If AI does not mention you in any of these tests, you have zero AI visibility. If it mentions you but with incorrect information, you have a data quality problem. Both are fixable.

The booteek B.E.S.T. Score Framework:

We built a scoring system specifically for restaurant and bar AI visibility. B.E.S.T. stands for Business Visibility, Employee Excellence, Service Quality, and Traction & Growth. Each component scores 0-25, for a total of 100. Most UK venues score below 30 on their first assessment.

The score tells you exactly where you are strong and where you are losing visibility. A venue with a high Business Visibility score but low Traction score knows their profile is complete but they are not generating enough fresh review content.

Want an instant assessment? Our Digital Health Check analyses your Google Business Profile completeness, review response rate, and AI visibility indicators in under 60 seconds. It costs GBP 29 and gives you a prioritised action list specific to your venue.

The Bar Blindspot: Why Pubs and Bars Are Invisible to AI

Every GEO guide, every AI visibility article, every SEO consultant talks exclusively about restaurants. Search for "how to get your restaurant found by AI" and you will find dozens of results. Search for "how to get your bar found by AI" and you will find almost nothing.

This is not because bars do not benefit from AI visibility. It is because nobody in the industry is talking about them. That is a blindspot — and it is an enormous opportunity for bar owners who act now.

Bar-specific queries have completely different patterns to restaurant queries. "Best cocktail bar Manchester" is a different search intent to "best restaurant Manchester." "Pub with beer garden near me" has different signals. "Where to watch the football tonight" is a query that AI answers using attributes most bars have never filled in.

The Google Business Profile attributes that matter for bars are different too. Happy hour, live music, cocktail menu, real ale, draught beer selection, late night opening, pub quiz nights, live sport, dog-friendly — these are the signals that AI uses to match bars with specific customer queries. Our audit found that 73% of UK bars have fewer than 5 attributes completed on their Google profile. The average restaurant has 8.

Here is the competitive reality: if you run a bar and you complete your Google profile, claim your Bing listing, and respond to reviews within 24 hours, you will likely be the only bar in your area doing all three. The competition for AI visibility among UK bars is effectively zero. The first movers will own it.

What NOT to Do: Common Mistakes That Kill AI Visibility

Knowing what to do is half the battle. Knowing what to avoid is the other half. These are the most common mistakes we see UK restaurant and bar owners make when trying to improve their AI visibility.

Keyword stuffing your Google profile description. A study by researchers at Princeton found that keyword stuffing actually reduces AI visibility by 10% (Princeton GEO Study, 2024). AI assistants are trained to detect unnatural language. Writing "Best Italian restaurant Manchester Italian food Manchester Northern Quarter Italian dining Manchester" will hurt you, not help you. Write naturally.

Uploading PDF menus and assuming AI can read them. It cannot. We covered this in Step 3, but it bears repeating: 98% of UK venues rely on PDF menus. Every single one is invisible to AI menu queries.

Ignoring Bing entirely. Most UK business owners think Google is the only search engine that matters. For traditional search, that is mostly true. For AI, it is dead wrong. Bing feeds ChatGPT, which has over 200 million weekly active users. Ignoring Bing means ignoring the largest AI assistant on the planet.

Writing generic review responses. "Thanks for your review!" tells AI nothing. It adds no structured data, no keywords, no context. Every review response is an opportunity to add information that AI can index. Use it.

Setting and forgetting. Your Google profile is not a one-time task. AI penalises stale information. Update your profile monthly at minimum — seasonal menus, new photos, updated hours, fresh Q&A entries. Freshness is one of the strongest AI ranking signals, and 76.4% of top-cited pages are updated within 30 days.