Something changed in how people find a place to eat, and most restaurants have not noticed yet.
A few years ago somebody typed wings near me and got a list of ten links. Now a decent share of people ask a full question out loud or type it into an assistant. Where can I get wings near Franklin that is open after ten. Somewhere for twelve people with a kid friendly menu. And instead of ten links they get one or two recommendations and a reason.
That is a different game. You are no longer competing to be on a list. You are competing to be the answer. Here is what decides it.
The machine has to be able to read you
This is the one that disqualifies the most restaurants, and it is not sophisticated. Your menu, your hours and your address have to exist on your website as actual text.
If your menu is a PDF, or worse a photo of a printed menu, an assistant mostly cannot tell what you sell. Same for hours baked into a graphic, or an address that only appears inside an image of your storefront. It is not that the machine refuses. It is that there is nothing there to read.
Put the menu on the page as text with prices and descriptions. Put the hours on the page as text. Keep the PDF for printing.
Answer the questions people actually ask
Voice and AI searches are questions, phrased the way a person talks. Do you take reservations. Is there parking. Do you have anything gluten free. Are you open on Thanksgiving. Can you do a party of fifteen.
If those questions and their answers appear in plain language on your site, you become the easy source for them. If they do not, the assistant guesses from a review or skips you for a place that said it clearly.
Write them the way a guest would say them, not the way a brochure would. Nobody asks about your accommodation capabilities for larger parties.
Be specific, because generic is unrecommendable
An assistant recommending a restaurant needs a reason. Specific facts are reasons. Vague atmosphere language is not.
Late night kitchen until midnight on weekends. Gluten free crust. Smash burgers. Dog friendly patio. Private room for twenty. A dozen specific true things about your place will get you recommended far more than three paragraphs about passion and quality ingredients, which every restaurant on earth also claims.
Label your information so a machine can file it
There is a standard format, called structured data, that puts labels on the facts on your page. This is the restaurant, these are the hours, this is the address, this is the menu, these are the reviews. It is invisible to guests and it removes all guesswork for anything reading the page.
Most restaurant sites do not have it, and almost every site builder can add it. If you are not going to touch code, this is the one thing worth asking whoever built your site to do.
Keep your details identical everywhere
Your website, your Google listing, Yelp, the delivery platforms, Facebook. Assistants cross-check these, and when they disagree they lose confidence in all of them and recommend somebody else.
Hours are the usual culprit, especially holidays and the night you started closing early in winter. Pick an afternoon, open every listing you have, and make them match exactly, down to the way you abbreviate your street.
Do not block the crawlers you want
Plenty of sites quietly tell AI crawlers to stay out, either because a plugin did it or because somebody read a scary article. If you want to be recommended by an assistant, let the thing read your site.
Worth checking with whoever runs your site, and worth getting right, because it is a one line fix that otherwise makes everything else here pointless.
Mentions elsewhere matter more than you think
These systems learn from the whole web, not just your site. A write-up in a local paper, a spot on a town events page, a food blogger post, a chamber listing. Those are the sources an assistant leans on when it decides which place to name.
You do not need a PR agency. You need to be in the handful of local places that write about restaurants in your county.
Test it tonight, it takes four minutes
Ask an assistant where to get your kind of food in your town, the way a stranger would. See if you come up. Then ask it directly about your restaurant by name and read what it says back.
If it does not know you exist, you have a reading problem, usually the menu or the structured data. If it knows you but gets the hours or the menu wrong, you have a consistency problem. Either way you just found your list, and it cost you nothing.
Next: Your menu prices are from last year. Your food cost is not.