Every other category we track has an ad auction under way, whichever way that money is being spent well or badly. London hospitality does not, not yet. Across 585 answers we tracked in London hospitality, wine bars and restaurants, from our own daily multi-engine tracking, which includes ChatGPT ad tracking, only 11 carried an ad: 1.9% coverage, the lowest of any category we measure. One restaurant is bidding. The rest of hospitality's paid inventory went to fashion, real estate, theatre tickets and hotel booking platforms with no connection to the query being asked. Meanwhile the organic side of this category is the most concentrated thing we track anywhere: Time Out alone is retrieved in about 64% of London ChatGPT dining answers. Hospitality is the mirror image of everything else in this research. Paid has not arrived. Organic already decided the answer.
TL;DR
- London hospitality carries ads in just 1.9% of tracked answers (11 of 585), the lowest ad coverage of any category we measure.
- Only one hospitality competitor is bidding: Paro, an Indian restaurant in Covent Garden, took 2 of the 11 ad slots, both on Canary Wharf dining queries outside its own postcode.
- The rest of the inventory is unrelated fill: fashion, real estate, theatre, hotel and travel booking, a bank card and a wholesale supplier, none of it built for a dining query.
- Meanwhile organic citation is fiercely concentrated: Time Out in about 64% of London ChatGPT dining answers, Eater 40.1% in New York, and Berlin's city tourism board 44.9% on ChatGPT.
How much of London hospitality search is actually monetised with ads?
1.9% is the lowest ad coverage of the five categories we track daily, a long way below waste and hauling software at 77.8% and even below UK trades at 3.6%.
| Category | Value |
|---|---|
| Waste, dumpster and hauling software | 77.8% |
| Entity resolution and identity data | 73% |
| Finance automation, treasury, AP and payroll | 7.1% |
| UK trades and gas compliance | 3.6% |
| London hospitality, wine bars and restaurants (highlighted) | 1.9% |
The timing pattern is as telling as the number itself. Our tracking recorded a single ad on 6 July, then nothing again until the final two days of the window, when ten more appeared at once. That is not a category that has been quietly building an ad market over weeks the way entity resolution has, with ads present continuously as far back as late May. It is a category almost nobody has tested yet, with a late, small burst of activity right at the end of our tracking window, the kind of pattern you would expect from one or two advertisers trying the inventory for the first time rather than an established market finding its rhythm.
Who is actually buying hospitality ad inventory?
We reviewed all 11 of the ad bearing answers in this category.
| Advertiser | Ad slots | What they actually sell |
|---|---|---|
| Paro (Covent Garden) | 2 | Indian restaurant, corporate and evening dining |
| Booking.com | 2 | Hotel and travel booking |
| Tesco F&F | 1 | Fashion retail |
| Double Puc Cafe | 1 | Cafe |
| Wild Science Lab | 1 | Unrelated to dining |
| Avenue Q | 1 | Theatre |
| Barclaycard | 1 | Credit card rewards |
| Empire Supplies | 1 | Wholesale and trade supplies |
| Trip.com | 1 | Travel booking |
| A real estate advertiser | 1 | Property |
| Category | Value |
|---|---|
| Paro (Covent Garden) (highlighted) | 2slots |
| Booking.com | 2slots |
| Tesco F&F | 1slots |
| Double Puc Cafe | 1slots |
| Wild Science Lab | 1slots |
| Avenue Q | 1slots |
| Barclaycard | 1slots |
| Empire Supplies | 1slots |
| Trip.com | 1slots |
| Real estate advertiser | 1slots |
Is any restaurant actually bidding on London dining queries?
Only one, and the detail is more interesting than the headline number suggests. Paro, an Indian restaurant based in Covent Garden, took 2 of the 11 slots we found, and both ads ran against dining queries for Canary Wharf, a different part of London entirely: "private dining room in Canary Wharf," "restaurant near Wood Wharf with great wine," "best restaurant in Canary Wharf for wine lovers" and "waterfront dining in Canary Wharf" were the prompts carrying ads in our sample. Paro's ads, "Corporate Dining, Covent Garden" and "Dinner in Covent Garden," are deliberately bidding outside its own neighbourhood to reach a Canary Wharf searcher, corporate dining and private events being exactly the kind of booking where a diner will travel across town for the right room. It is the single clearest example in this entire dataset of a hospitality business using AI search ads the way a software vendor would: bidding on a competitive, high intent query rather than waiting to be found.
What is filling the other ad slots instead?
Almost everything else in the sample has no real connection to the query it appeared against. Booking.com and Trip.com sell hotel and travel bookings, adjacent to hospitality in the broadest sense but not competing for a restaurant table. Tesco F&F is a fashion retail line, Avenue Q is a West End theatre show, Barclaycard is a bank card rewards programme, Empire Supplies is a wholesale trade supplier, and a real estate advertiser rounds out the list. Double Puc Cafe and Wild Science Lab are the two hardest to place, small operators picked up by whatever network fill an AI engine's ad system had available at the time. None of this reads as deliberate targeting by a hospitality competitor the way Paro's two placements do. It reads as unclaimed inventory being filled by whoever happened to be in the auction.
Meanwhile, how concentrated is the organic side?
This is the mirror image of the paid story, and it is the real headline for anyone running a restaurant or wine bar. While almost nobody is buying ads, a small number of sources are deciding almost everything about who gets recommended. In our own published research on which sources ChatGPT cites for London dining, Time Out was retrieved in about 64% of London ChatGPT dining answers, a single editorial title deciding the shape of most answers. The same pattern repeats city by city. In our New York dining study, Eater was retrieved in 40.1% of answers, Time Out 36.9%, The Infatuation 35.7%, Reddit 28.4%, and the booking platforms Resy and OpenTable in 25.4% and 24.0%, across 830 answers and 24 restaurants. In our Berlin index, visitberlin.de, the city's own tourism board, was retrieved in 24.9% of all answers and 44.9% of ChatGPT answers specifically, with Reddit playing a far smaller role at just 10.1%, across 823 answers and 22 restaurants.
| Metric | Value | Caption |
|---|---|---|
| London, ChatGPT | ~64% | Time Out |
| New York, all engines | 40.1% | Eater, most retrieved source, 830 answers |
| Berlin, ChatGPT | 44.9% | visitberlin.de, the city tourism board |
Three cities, three different leading sources, editorial in London, a mix of editorial and booking platforms in New York, a city tourism board in Berlin, and in every one of them a small handful of sources carries most of the answer. For the fuller picture of London specifically, including which restaurant groups already lead the index, see our UK Restaurant AI Visibility Index, and for how the platforms selling AI visibility measurement to restaurants stack up against each other, our comparison of the major vendors is worth reading alongside this one.
How fast could a hospitality ad market actually arrive?
Faster than the current numbers suggest, if the other categories we track are any guide. UK trades and gas compliance carried zero ads across the first two weeks of our tracking window, nothing at all, and then went from a first placement on 7 July to roughly half the entire month's ad volume landing in the final three days. A category can sit at effectively nothing for weeks and then fill in within days once a handful of advertisers decide the inventory is worth testing. Hospitality's own pattern, one ad on 6 July and then ten more in the final two days of our window, looks like the very early version of exactly that curve. Paro found this inventory before any other hospitality operator did. There is no reason to assume it stays alone for long.
What does this mean for restaurant groups and wine bars right now?
The fight in hospitality is still entirely organic, and it is entirely about citations: which sources an AI engine trusts enough to pull from, and whether your restaurant is actually present in them. Paid has not arrived in any meaningful way, one restaurant and a scattering of unrelated fill is not an ad market. This is part of a wider pattern we track across which categories ChatGPT actually monetises, and hospitality sits at the opposite end of it from the two B2B categories we cover alongside this one: in finance automation, a lead generation aggregator is out-advertising the real software competitors, and in team collaboration software, almost nobody selling the actual product is bidding at all. Hospitality has the opposite gap, empty paid inventory rather than badly contested paid inventory, but the underlying lesson is the same: whoever moves first gets the cheapest entry. Every other category we track eventually developed some level of ad activity, and hospitality's own tracking already shows a small burst of new placements in the final two days of our window. The organic groundwork, being present in the handful of sources that actually build the answer, is the part worth securing now, while paid is still this quiet.
Disclosure: this research comes from our own daily multi-engine tracking, which includes ChatGPT ad tracking, run continuously across London hospitality and four other B2B and consumer categories. We are Schmitdy, AI Heroes' own AI search growth engineering service, and we publish this because nobody else is tracking which categories AI search actually monetises and which ones are still wide open. The free AI search audit shows where your own restaurant or group sits against this picture.
