Many people ask how ChatGPT recommends a restaurant, but the answer really lies in accurate data and technical settings. Anyone working in hospitality can see that AI assistants play a growing part in guests' decisions. ChatGPT, Perplexity and Gemini give direct answers, and users like that speed. For a restaurant this is a new source of traffic, and making use of it takes serious technical work, because the machines decide on the basis of algorithms and structured datasets. The website has to speak this particular machine language, which the industry calls AEO.

Why AI recommendations are worth the effort

When a potential guest asks an AI for the best places, the assistant weighs the information available on the web within seconds. Our job is to get the restaurant to the very top of the results. Structured data and precise location data help the system find its way, because the bots mainly read text, data written into the code and databases, not atmospheric photos. For a high-traffic Budapest restaurant, we started by putting the technical foundations in order. Over the same period, the number of daily sessions coming from ChatGPT grew from roughly 25 to 50 between early summer and early September. Tourists often ask their phones for recommendations in their own language, and the system builds a list from local data.

Search engine optimisation now extends to AI-based systems too, and the emphasis is clearly on machine readability. The most important lesson is that AI recommends largely on the basis of location data and the machine-readable data on the website. What matters most is the accuracy of the Google Business Profile, the map data and guest reviews, and that they match perfectly. If the data contradict each other, the machine becomes uncertain and suggests another place instead.

Measuring visibility in practice

Measuring traffic accurately is the first step towards a clear picture and deliberate marketing. With GA4 it is now possible to track very precisely how many visitors reach a restaurant's website from AI assistants. These modern platforms show up clearly among the referring sources. At one of our clients, visits from AI assistants made up only 1.0% of all website traffic in May 2026. After measurement was introduced and the technical foundations were put in order, the figure stood at 6.8% at the end of September 2026.

Besides structuring the data, AI search crawlers were deliberately allowed onto the website, which is a very useful technical step. An llms.txt file prepared for the bots helped the assistants process the information quickly. Setting up Search Console and Bing Webmaster Tools properly was also part of the technical foundations. A flawless technical background is the basis of trust in the eyes of the machines.

AI recommends largely on the basis of location data and the machine-readable data on the website, so the accuracy of the Google Business Profile, the map and the reviews matters most.

We also checked the effect of the technical work in a live test, to see the real user experience. In a search on 6 September 2026, ChatGPT recommended the restaurant in question in first place in its category. Next to the result, a very detailed place card appeared, showing the address, opening hours and current ratings at once. The place card is the most eye-catching part of the answer, so the data on it must be flawless.

How bookings and revenue developed

Growth in traffic alone means little to a restaurant, because at the end of the day conversion is the real measure. Of the 1,037 table bookings made on the website in September 2026, 85 came specifically from AI assistants. That is 8.2% of all online bookings, a clearly measurable result that makes business sense. Based on 90 days of data, we also found that 98% of AI-sourced bookings came directly from ChatGPT.

The quality of conversion rates is also telling when the different traffic sources are compared. We examined the figures on the website over a two-week period, from 20 August to 2 September 2026. According to the data, visitors from ChatGPT booked slightly more often than those from Google (4.3% versus 3.7%), and more often than the site average (3.1%). We estimated the financial effect of the bookings using the average bill value from the till system. On that basis, this channel brings in more than HUF 3 million in revenue per month. Cancellations and no-shows were not deducted, but the estimate shows well how much revenue comes from this channel today.

Data integration and the technical background

Optimising for AI search takes precise technical work, best left to an experienced developer. Without the right foundations, assistants simply cannot find the restaurant amid the vast digital noise. The machine-readable data sheet, that is, the structured data, plays a key role on the website, because this code tells the machine the most important parameters: exactly where the place is, when it is open and which dishes are on the current menu. Listing menu items, prices and allergens accurately helps machine understanding a great deal.

Putting the Google Business Profile in order is a basic task that has to be done at the start of any optimisation. Many AI systems pull place card data directly from it when answering. If the details on the website and on the Business Profile differ, the machine becomes uncertain, and uncertainty immediately reduces the chance of appearing. During the SEO clean-up we remove these annoying contradictions, and our AI Ready package contains exactly these steps. Reviews and ratings also strongly shape machine recommendations, since AI search tools constantly read guest feedback. Positive, detailed reviews significantly strengthen a place's position in the digital space, so quality hospitality pays off online as well.

The role of content in AEO

Besides the technical settings, the quality of the written content also carries a lot of weight in ranking. The texts on a website have to be completely unambiguous so that bots can easily extract the essentials. Complicated wording only makes the machines' job harder when crawling. Relatively short, clear statements and logical paragraphs help the information to be processed quickly. Menu descriptions should always detail the ingredients and cooking methods, because machines use these small details too when giving accurate answers. The particular nature of tourism in Budapest makes AI visibility especially important, since foreign guests rely almost entirely on their phones to find their way around.

Technical SEO clean-up is not a one-off task, of course, but the start of a long-term process in a marketing strategy. The algorithms of search engines and AI models are constantly updated, which calls for continuous attention and data maintenance. Search Console data show how Google sees the website.

What to do now

The share of AI-based searches keeps growing in hospitality too. Falling behind competitors technically can cause a serious loss of traffic. With the right steps, however, any restaurant can be prepared for this new era.

  • Check the Google Business Profile details and make sure the name, address and opening hours match those on the website.
  • Ask a developer to add structured data, because this is what machines use to understand a menu most easily.
  • Measure traffic in GA4 and separate visitors from AI assistants to see the real proportions.
  • Allow AI search tools to access the website; an llms.txt file helps the bots a lot.
  • Write rich and accurate descriptions of the dishes, because machines give the most relevant recommendations based on written information.