Being named by an AI engine is downstream of being in the sources it reads. So the most useful thing a London restaurant can know is not a vague list of best practices, it is exactly which sources ChatGPT and the other engines actually pull when they answer a dining question. We measured it: across two weeks in July 2026 we tracked which source domains the major AI engines retrieved and cited for around 85 real London dining questions, run daily. The picture is concentrated, it differs by engine, and it hands you an editorial map. This is a companion to the UK Restaurant AI Visibility Index; the index measured who gets named, this measures what decides it.
TL;DR
- On ChatGPT specifically, Time Out was retrieved in about 64% of London dining answers, nearly two in three. It is the single most influential source in the city's AI dining answers.
- The full landscape is a small set: after Time Out come the community listings site thatsup, The Infatuation, Reddit, SquareMeal, Michelin, OpenTable and a handful of others.
- Being retrieved is not the same as being cited. Reddit and Michelin are quoted far more heavily when they appear than a listings lookup like DesignMyNight, which gets pulled for data but rarely quoted.
- The practical read: this is your editorial target list. Presence in these specific sources is what puts a London restaurant into AI answers.
How we measured it
We ran around 85 real London dining questions, best restaurants in a neighbourhood, a wine bar for a date, a private room for a number of guests, daily across the major AI engines for two weeks in July 2026. For every answer we recorded which source domains the engine retrieved, and how heavily each was cited inline. That gives two different measures worth separating: how often a source is retrieved, meaning pulled into the working set for an answer, and how heavily it is cited, meaning actually quoted in what the diner reads. It is a young, two-week dataset and a first read, refreshed quarterly, not a definitive census.
Which sources does ChatGPT cite for London restaurants?
A small, concentrated set, led overwhelmingly by one name. On ChatGPT, these were the most-retrieved sources for London dining answers.
| Source | Type | Retrieved in (ChatGPT) |
|---|---|---|
| Time Out | Editorial | 64% |
| thatsup | Community listings | 56% |
| The Infatuation | Editorial | 42% |
| Community | 33% | |
| SquareMeal | Editorial | 32% |
| Visit London | Institutional | 30% |
| OpenTable | Booking | 29% |
| Condé Nast Traveller | Editorial | 28% |
| Michelin | Guide | 26% |
| DesignMyNight | Listings | 25% |
Time Out is not just first, it is dominant: it was pulled into nearly two thirds of ChatGPT's London dining answers. For a restaurant, a single well-placed Time Out listing is worth more in AI visibility terms than almost any amount of work on your own website. Across all the engines combined its share is lower, around 42%, but it still tops the table, which tells you the editorial concentration is a feature of the whole landscape, not one engine.
Retrieved is not the same as cited
This is the subtlety most source lists miss, and it changes where you spend effort. Some sources are retrieved often and quoted heavily; others are pulled for a quick fact and barely quoted. In our data, Reddit and Michelin had the highest citation intensity: when an engine retrieved them, it leaned on them hard, quoting them well above average. A listings site like DesignMyNight, by contrast, was retrieved frequently but cited very lightly, the engine checks it for a detail rather than repeating its verdict. The lesson: prioritise the sources that are both widely retrieved and heavily cited, because those are the ones whose words actually reach the diner. Reddit's weight here echoes what we found in Reddit and local forums behind restaurant AI answers.
The source landscape, by type
Grouped by kind, the London dining answer is built from four buckets, and you need presence across more than one.
| Source type | Examples | Role in the answer |
|---|---|---|
| Editorial | Time Out, The Infatuation, SquareMeal, Condé Nast Traveller | The backbone; sets the shortlist |
| Community | Reddit, thatsup | Trust and consensus; heavily cited when pulled |
| Listings and booking | OpenTable, DesignMyNight, Michelin | Facts, availability and credentials |
| Institutional | Visit London | Broad, safe reference for tourist-facing queries |
No single source carries you, and the mix shifts with the question: a tourist-facing query leans more on institutional and editorial sources, while a locals' where-to-eat-tonight leans harder on community. The engines vary too, so being present in only one place leaves you exposed to how any one model weights its sources.
What to do with this: your editorial target list
This is the payoff. Instead of guessing where to seek coverage, you have a ranked list of the sources that actually feed London AI answers. In rough priority: earn a place in Time Out and The Infatuation, get onto SquareMeal and the relevant city editorial, be genuinely and honestly present in the Reddit and community threads for your area, and make sure your listings and booking entries, OpenTable and the like, are complete and consistent. That is not a generic PR plan, it is the specific map the data draws. For how this editorial and citation work sits alongside the technical fixes, and who should own it, see who actually fixes AI search visibility.
Disclosure: Schmitdy is our own AI search service at AI Heroes, and this study comes from the same daily multi-engine tracking we run for the restaurants we work with. Pursuing placement in exactly these sources, and tracking whether it moves your citations, is a core part of what we do. If you want to see which of these sources currently name you, and which name your competitors instead, the free AI search audit shows the gap.
