The decisive asset in restaurant search is the menu, and most restaurants publish it in the least useful possible form. A menu that is an image or a downloadable file cannot be read by a search engine, cannot be read by an assistant, cannot be read by a screen reader, and cannot be searched for a dish. Meanwhile dish-level intent is a genuine and growing query pattern: people search for the specific thing they want to eat tonight, then find the restaurant that serves it. A restaurant with an unreadable menu is invisible to every one of those searches.
Reservations are the second structural issue. Booking platforms are useful, and they also insert themselves between you and the guest, own the diner relationship, and in some markets sell your covers alongside competitors' on the same screen. There is no reason to abandon them, but a restaurant that never builds a direct reservation habit is renting its own regulars. The campaign's job is to make the direct path the natural one for people who already know the name, while the platforms keep doing discovery work.
Then there is the part of the P&L that quietly decides the year: private dining, buyouts, holiday parties and rehearsal dinners. That demand arrives from a planner rather than a diner, uses different search language entirely, needs capacities, minimums and menus in advance, and books weeks or months out. It is also the least competitive search space most restaurants have access to, because so few of them build for it. Review velocity, sentiment and photo freshness underpin all of it, since a restaurant's reputation is re-litigated in public every single week.