Why Some of Your Restaurants Show Up in AI Answers and Others Don't (2026)

TLDR: If one of your restaurants appears in AI answers and a sister location doesn't, the shared website is rarely the cause. Assistants judge each location on its own evidence: a specific category, named dishes, local guides and press, recent reviews. Compare the sources listed for your strongest and weakest locations, then build what the weak one lacks.

You ask an assistant for a good brunch near your newer restaurant. It names two places on your street and then, third, your original restaurant on the other side of town. Same group, same website, same standard of cooking. As far as the answer is concerned, one of your locations exists and the other does not.

A common reaction is to brief a website fix. We think that is the wrong place to start. The assistant clearly knows your group: it just named your flagship. What it lacks is evidence about the newer location, and most of that evidence lives outside your website.

Why does AI name one of your restaurants and not the other?

Because assistants answer place by place. When a guest asks for brunch in a neighborhood, the engine looks for locations that match the question, sit near the asker and are well known, then builds its answer from whatever it can find about each one. Your flagship's history does not carry over to the newer location.

Google states these three tests plainly for its local results. Its local ranking guidance names relevance, distance and prominence, says businesses with complete and accurate information are more likely to show up, and bases prominence partly on how many websites link to a business and how many reviews it has. Its category help adds that the categories you select affect your local ranking. For AI Overviews and AI Mode, Google says there are no additional requirements, and it recommends keeping Business Profile information up to date.

Other assistants publish less, but the measured evidence points the same way.

Is it just the number of reviews?

Not on its own, although this is where most groups look first. In a study of 10,000 US restaurants published in June 2026, with data from May, Local Falcon found 74.9% invisible in Google's AI Overviews recommendations, against 19.6% on Google Maps. Restaurants with more than 1,000 Google reviews were still left out 70.9% of the time. The authors' summary: "Review volume barely moves the result." Local Falcon sells local search software, and the study tracks associations, not causes, but it is a large sample.

An independent audit gives reviews more weight. Vladimir Pitenin's August 2026 preprint counted every cafe, restaurant and bar in two areas of Bali, 4,776 venues, and collected 2,208 search-grounded answers to 96 guest-style questions from ChatGPT, Claude, Gemini and Perplexity. 85.6% of venues were never recommended. Having a website was associated with higher odds of being recommended (odds ratio 1.92), as were review volume (1.64), listed prices (1.54) and mentions on other websites (1.44). Star rating had no measurable effect on getting in at all; it only helped decide which recommended venue came first.

The two studies differ on review volume, and neither shows cause. For an operator, the safe reading is that volume is necessary but not sufficient: a location with a few dozen reviews starts behind, but adding more will not close the gap if nothing else is said about it. What the reviews say, and what other sites say, is where the weaker location usually falls short. Our guide to whether Google reviews affect AI restaurant recommendations covers that evidence in full.

So what does the strong location usually have?

More proof, from more places, of what it is good for. A flagship that has been open for years tends to have a specific category on its profile, dishes that guests name in reviews and photos, a write-up in the local paper, a place on two or three lists of the best brunch spots nearby and a booking page that matches its listing. A newer sister location often has a generic category, a few dozen reviews and nothing written about it elsewhere.

That second gap matters because AI search leans on other people's pages. A 2025 study of AI search engines by Chen and colleagues found "a systematic and overwhelming bias towards Earned media", meaning third-party sources rather than a brand's own pages.

How do you find the gap in one hour?

You cannot see an assistant's weights, but you can see what it says and which sources it lists beside the answer. Several assistants show them: Perplexity adds numbered citations to each answer, and Google shows supporting links in AI Overviews and AI Mode. Those lists are your evidence. They show what was available about each location; they do not prove which page the engine relied on.

Pick two locations of similar standing, your most visible and your least visible, and run this test in the assistant your guests are most likely to use:

1. Write ten guest questions per location. Use the occasions each location is built for, such as brunch with a stroller or a quiet dinner before the theater, and name the neighborhood or street in every question. Distance is one of the three tests, so a question without a place in it measures where you are sitting, not your location.
2. Ask the same questions about both locations on the same day. Answers vary between runs and between assistants, so the comparison only holds when the conditions match.
3. Record three things per answer: whether your location is named, its position in the list, and every source listed alongside the answer.
4. Open each listed source and note what kind it is: your website, a listing profile, a review site, a guide or list, local press, social media or a competitor's page.
5. Tally the sources by type for each location. The difference between the two tallies is your gap.

The arithmetic: 20 questions at about two minutes each to ask, read and log is 40 minutes; opening and sorting the sources takes most of the remaining 20. If you have a second hour, repeat the questions in a second assistant.

What does the gap usually look like?

Once the sources are sorted, the gap falls into six evidence types. The table below is the sheet we suggest you fill in. The middle columns show what a strong and a weak location typically record; replace them with your own notes.

In a typical result, your website appears in the sources for both locations. The difference is everything around it.

When is the website the problem after all?

When the weaker location has no page of its own. In the Bali audit, having a website was the strongest single signal measured. If the newer location shares one locations page with the others, sends guests to a PDF menu, or has no booking link of its own, an answer has little to quote from your site about it. Fix that first: our guide to how restaurant groups help guests find the right branch sets out what each location page needs, and what local SEO still does for multi-location restaurants covers the listings work behind it.

The second exception is age. A location opened in the last few months starts at zero on every review site and in every guide, whatever the group's reputation. That is a matter of time, not neglect, and our guide to how restaurant groups protect AI visibility while expanding covers the routine for new sites.

If neither applies, both locations have complete pages and both have been open for a year or more, the gap is outside your website, and another site rebuild will not close it.

What should you fix first for the weaker restaurant?

Work from the cheapest and fastest fix to the slowest, because the early fixes are entirely in your control and the later ones depend on other people:

1. Make the core facts match everywhere. Profile, location page and booking site should agree on address, hours and booking link; Google lets you edit a verified profile to keep it accurate. Our free restaurant listing mismatch checker compares your website with public listings.
2. Set a specific category and name the dishes. Google's guidelines ask for categories as specific as possible but representative of the main business. Name the dishes the location is known for on the profile and on a text menu on its page, and mark them up: Google's local business structured data and the schema.org Restaurant type both have fields for the cuisine served and the menu.
3. Earn outside mentions for what the location does best. Pitch the local paper and the writers behind the lists that appeared in your test, with the one thing this location does that the flagship doesn't. This takes months, so start now.
4. Ask every guest for a review at one fixed moment. You want recent reviews that name what the location is good for. Our guide to restaurant review management for multi-site groups covers replies and the rules on asking.
5. Re-run the same ten questions in four weeks. Same assistant, same wording, both locations. Fact and category fixes wait only on platforms updating; mentions and reviews take longer to build.

This comparison is where our free AI Search audit starts for a restaurant group: it runs the guest questions location by location across ChatGPT, Claude and Google AI, lists the sources shown with each answer and turns the weaker location's gaps into a 30-day plan for each team. Where we then take on the work, Google Business Profile optimization and menu pages are included at no extra charge.

What should you do this week?

Fill in one sheet. Take your strongest location and your weakest, write ten questions for each, and sort the sources you find into the six rows above. By the end of the hour you will know whether the weaker location needs a page, a category, its dishes named or a reason for someone else to write about it. Fix the facts and the category first, because they are yours to change today, then run the same questions again in four weeks.

If you would rather we ran the comparison, request a free AI Search audit and name the two locations you want compared.

Sources

1. Google Business Profile Help: Tips to improve your local ranking on Google
2. Google Business Profile Help: Edit your business category
3. Google Business Profile Help: Guidelines for representing your business on Google
4. Google Business Profile Help: Find and edit popular dishes for restaurants
5. Google Search Central: AI features and your website
6. Google Search Central: Local business structured data
7. Schema.org: Restaurant type
8. Perplexity Help Center: How does Perplexity work?
9. Pitenin, Invisible to the Machine: Auditing AI Restaurant, Cafe, and Bar Recommendation Against a Complete Market Census (arXiv:2608.07069, preprint, August 2026)
10. Chen, Wang, Chen and Koudas, Generative Engine Optimization: How to Dominate AI Search (arXiv:2509.08919, preprint, September 2025)
11. Google Business Profile Help: Edit your Business Profile
12. Local Falcon: The Restaurant AI Visibility Index (vendor study, published June 2026, data from May 2026)