A restaurant that is a household name at home opens in a new city and discovers something its marketing budget did not warn it about: to an AI engine, it is a stranger. Brand recognition is a human thing. When a diner in the new city asks ChatGPT or Google's AI where to eat, the model does not reason about your reputation elsewhere. It looks for a venue it can place in this city, with local reviews, local citations and a resolved local identity, and finding none, it names the incumbents it is already sure of. Expansion is the single moment AI visibility matters most, because you are spending the most and starting from zero, and it is the moment almost no one does the work.
TL;DR
- A famous name earns little in AI answers in a new city. An engine needs a resolved local entity plus local corroboration, none of which your reputation elsewhere provides.
- A new location starts with no local reviews, no local best-of citations and no community footprint, so it sits below the confidence bar an engine needs before it will recommend you.
- The fix is a deliberate cold-start sequence: entity, schema and listings live before you open, editorial and community outreach timed to launch, and early reviews and area-anchored content in the first weeks.
- For a group, every new venue is its own cold start, no matter how strong the brand, which is where multi-venue plans usually underinvest.
Why does a new location start invisible even with a famous name?
Because the engine is answering a local question with local evidence, and you have not created any yet. When someone asks for the best place to eat in a specific new-city neighbourhood, the model assembles its answer from venues it can resolve to that place and corroborate through reviews and cited sources. Your acclaim in another city is not local evidence. There is no local review history, no local roundup has named you, and your entity has not been disambiguated to the new address. So the safe move, the one an engine is built to make, is to name a venue it can stand behind and leave the newcomer out. This is the same trust threshold we describe in why your restaurant is invisible on ChatGPT, met at its hardest: from a standing start.
The cold-start problem, concretely
Four things are missing on opening day, and each is a reason to be skipped.
| What the new venue has | What an AI engine needs to recommend it locally |
|---|---|
| A strong brand from another city | A local entity resolved to the new address and area |
| A launch and a website | Local reviews with some depth and recency |
| Press about the brand generally | Local "best of" and where-to-eat citations for the new city |
| An opening-week buzz | Community mentions in local forums and threads |
None of these arrive on their own quickly, and the gap between opening and being recommendable can run for months if nobody closes it deliberately. Worse, the star rating trap bites hardest here: a brand-new venue is the newest, thinnest-history entity of all, which we unpacked in the star rating trap.
The new-city cold-start playbook
You close the gap by front-loading the signals an engine needs, starting before the doors open.
| Phase | What to do |
|---|---|
| Before opening | Publish the new venue as its own entity with per-location Restaurant schema, get the Google Business Profile and every booking and listing platform live and consistent, and publish an area-anchored page that names the neighbourhood plainly |
| At launch | Time local editorial and where-to-eat outreach to the opening so the new-city roundups name you while the news is fresh, and make the menu crawlable as real text from day one |
| First 90 days | Earn early local reviews on the platforms engines read, show up honestly in local community and forum threads, and keep publishing area and occasion content so the local entity accumulates corroboration |
The point is sequence. An engine will not recommend what it cannot yet verify, so the work that creates local proof has to lead the opening, not trail it by a quarter. Getting into the local editorial roundups is the highest-leverage single move, because one citation in the new city can feed months of answers, the same dynamic we describe in AEO vs GEO vs SEO.
For groups, every new venue is its own cold start
A group expanding into a new city carries brand equity that helps a human recognise the name and does almost nothing for a model that has no verified local entity to attach it to. Each new site needs its own distinct entity, its own local reviews and its own local citations, exactly the entity discipline we set out in why your restaurant group's venues cannibalise each other. Groups tend to assume the brand will carry a new location in AI answers the way it carries footfall. It will not, and the venues that plan for a cold start at each opening pull ahead of the ones that assume the name is enough.
Is this relevant to UK restaurants?
Very, in both directions. UK groups expanding abroad, and into new UK cities, hit exactly this cold start, while the demand that makes it costly is already here: CGA by NIQ found 26% of UK consumers use AI apps to help decide where to eat or drink. The invisibility evidence, that most restaurants are absent from AI answers, comes from US audits by Local Falcon and is directional rather than a verified UK figure, but the cold-start mechanics are structural and hit any new opening in any market. A new venue that is invisible in AI answers is invisible to the growing share of diners who ask an engine first, in the very city where you most need to fill tables fast.
Disclosure: Schmitdy is our own AI search service at AI Heroes, so read this as the maker's case rather than neutral advice. Closing an AI cold start is a lot of what we do for new openings: standing up the local entity and schema, timing editorial and community work to launch, and tracking the major engines daily so you can see the new venue move from unnamed to recommended. For who should own the ongoing parts, see who actually fixes AI search visibility. If you have an opening coming and want to know where the new site stands in AI answers today, the free AI search audit shows the gap before launch.
