Diners who use AI to choose a pub or restaurant ask it for reviews first, prices second and the menu third. Zonal's latest GO Technology report, produced with NIQ across more than 5,000 British consumers, puts 38% of those AI users on menus, with opening times at 35% and deals and discounts at 28%. Reviews and ratings lead on 45%, average food and drink prices on 43%.
Nineteen per cent of consumers now reach for a tool like ChatGPT or a search engine summary when deciding where to go. A further 50% told Zonal they were considering it.
Four of the five requests are answered from the venue's own pages
A review sits with whoever collected it. The other four items on that list — prices, the menu itself, opening times, whatever offer is running — are facts the restaurant sets and publishes, and nobody else originates them. When a diner asks an assistant what a main costs at a particular pub, the model is either reading something the pub published or reconstructing it from a page that quoted it second-hand.
That puts the 43% on prices and the 38% on menus alongside the 35% on opening times, a field operators have maintained deliberately for years because a wrong closing time produces a complaint the same evening. Dish names and prices produce nothing so immediate, so they tend to live wherever the last redesign left them, often in a form built for somebody holding a phone rather than for anything parsing a page.
Making those four items machine-readable is a publishing decision rather than a marketing one. Menu software that emits Schema.org as a by-product gets there, GMMO among it, and so does hand-written JSON-LD that somebody remembers to revisit when the kitchen changes something.
AI sits six points behind the venue website
Zonal measured the older channels in the same survey. Recommendations from friends and family are used by 47% of consumers, online reviews by 32%, and venue websites by 25%. AI is on 19%.
Read those four figures as one list and the ordering is the finding. AI has come within six points of the venue website as a way of deciding where to eat, and a quarter of consumers still go to the site directly. Both routes are being asked the same questions about price, dish and opening time. Only one of them answers in the operator's own words.
The diners using AI are the ones spending most
The report puts 23% of guests who typically spend more than £60 a head on AI tools, against 14% of those spending under £5. Use among Gen Z runs at 34%, against 19% across all ages. In London it reaches 39%, against 16% for the rest of the UK. Consumers eating out more than three times a week were 42% more likely to use AI than those going out once or twice. None of those cuts describes an average customer, and a high-cheque site in central London with a young crowd is being researched this way far more often than the national figure suggests.
Trust is running well ahead of checking
Seventy-six per cent of respondents entirely or mostly trust AI engines to find a pub, bar or restaurant, and 72% say the same of AI-generated search summaries. Tim Chapman, Zonal's chief commercial officer, put the headline finding as "AI is rapidly becoming part of the customer journey." Karl Chessell, NIQ's director of hospitality operators and food for EMEA, named the other side of it: accuracy, prioritisation and protecting brand reputation are the challenges he set against the opportunity.
A diner who trusts the answer at those rates does not open the menu page to check it.
Chessell's accuracy point is an operational one, and it needs nobody to behave badly. A price goes up, the card on the table is reset to match, and the version a model took from the web weeks earlier is still the version it recites. Nothing in the exchange flags which of the two figures is current.
Accuracy here is not a question of good faith. It is whether the price being charged this week, and the dish list the kitchen is actually cooking, exist somewhere a machine can read them on the day the question gets asked — which is a property of how the menu is published rather than of how well the place is run.
