Restaurant SEO: How to Fill Tables From Google Maps and AI Search

Most restaurants do not lose bookings because the food is average. They lose them in the ninety seconds between “where should we eat tonight?” and someone tapping a competitor’s phone number. Restaurant SEO is the work of making sure your dining room wins that ninety seconds — on Google Maps, in the organic results, and now inside AI answers that recommend three places before a diner ever sees a search page.

Why restaurant search behaves differently

Most local industries deal with a considered purchase. A homeowner researching a furnace replacement will read three pages, compare quotes and call two days later. Restaurant search is the opposite: it is impulsive, mobile, time-boxed and heavily influenced by photographs. The average diner query happens within two hours of the meal and within a few kilometres of where they are standing.

That compresses everything. There is no nurture sequence. Either your listing appears with an appetising photo, a 4.6-plus rating, an accurate “Open now” status and a working reservation link, or the table goes to the restaurant that did.

Those four numbers dictate the strategy. Mobile-first means speed and thumb-sized tap targets. Review-driven means reputation is a ranking input, not a soft brand exercise. Map-pack dominance means your Google Business Profile is more valuable than your homepage. And PDF menus are the single most common self-inflicted wound in the industry.

The map pack is your front door

Google ranks the local pack on three broad factors: relevance, distance and prominence. You cannot move your building, so the work concentrates on the other two.

Relevance: describe what you actually sell

Your primary category is the highest-leverage field in the entire profile. “Restaurant” is a weak choice. “Neapolitan pizza restaurant”, “Vietnamese restaurant”, “brunch restaurant” or “halal restaurant” line up with how people search and with the refinement chips Google shows above the map. Add secondary categories for genuinely distinct revenue lines — catering, bar, bakery — but never stack unrelated ones to farm keywords; it dilutes relevance and invites suspension.

Fill every attribute Google offers. Outdoor seating, accepts reservations, good for groups, wheelchair accessible entrance, serves vegan options, late-night food, kid-friendly. Each one is a filter a diner can toggle, and each toggle removes every restaurant that left the field blank.

Prominence: reviews, photos, and the web talking about you

Prominence is Google’s read on how well known and well regarded you are. Three inputs move it fastest:

The posting habit almost nobody keeps

Google Posts expire, which is exactly why they work — they are a freshness signal you control. A weekly post covering the specials board, a seasonal menu change, a live music night or a holiday closure takes six minutes and keeps the profile visibly active. Restaurants that post weekly for three months consistently see higher discovery impressions than those that post once a quarter.

Your website still decides whether the click converts

The map pack gets attention; the site closes it. Four fixes cover the majority of lost bookings.

1. Kill the PDF menu

A PDF cannot be crawled usefully, cannot be read comfortably on a phone, and cannot rank for the dish names people search. Publish the menu as HTML: section headings, item names as text, prices as text, short descriptions, and dietary tags. Then add Menu structured data so search engines can pull individual items into rich results.

2. Make reservations one tap

Every path to booking should be reachable without scrolling: a sticky call button, an OpenTable or Resy link, and — if you take walk-ins only — a clearly stated wait-time expectation. Every extra tap between intent and confirmation costs roughly a fifth of the people who started.

3. Get under two seconds on mobile

Food photography is heavy. Serve it as WebP, size it correctly for the container, lazy-load everything below the fold and drop the autoplaying background video. A restaurant homepage should not weigh more than about 1.2 MB in total.

4. Write the pages diners actually search for

Structured data that earns rich results

Restaurants have one of the richest schema vocabularies available, and most use none of it. The full stack worth implementing:

• Restaurant — name, address, geo coordinates, phone, price range, cuisine served, accepted payments and opening hours specification, including holiday overrides.
• Menu / MenuSection / MenuItem — item names, descriptions, prices and dietary suitability. This is what feeds dish-level rich results.
• Reservation action — a potential action pointing at your booking platform so Google can surface a reserve button.
• AggregateRating and Review — only for reviews genuinely hosted on your own site; never mirror Google reviews into markup.
• Event — live music, tasting dinners, holiday sittings. Events are chronically under-used and frequently surface in dedicated modules.
• FAQPage — parking, dress code, corkage, group size limits, accessibility.

Keep hours accurate above everything else. A profile that says open when the door is locked generates the specific one-star review that costs the most: the one from someone who never ate the food.

Reviews: a system, not a hope

Review count and rating correlate with local pack position more tightly in food service than almost any other vertical. Treat it operationally.

1. Ask at the peak Train staff to ask when the plates are cleared and the guest has just said something positive — not on the way out the door.
2. Remove friction A QR code on the bill folder linking directly to the review form. One scan, one tap, done.
3. Reply to everything within 48 hours Replies are public and are read by future diners far more often than the review itself.
4. Handle negatives in a specific pattern Acknowledge, apologise for the experience, state the concrete change, invite them back offline. Never argue about the facts in public.
5. Track velocity monthly Ten reviews a month at 4.7 beats a hundred reviews in one week followed by nothing.

Photography is a ranking factor in practice

Google counts photo views, and diners choose with their eyes. A working cadence: eight to twelve new images per month, shot on a phone in daylight near a window, no filters. Cover the dishes people order most, the room at service, the exterior at the exact angle someone approaches from, and the staff. Geotagging images does nothing for rankings — the filename, the surrounding page content and the alt text do.

Seasonality: plan campaigns, do not react to them

Publish six to eight weeks ahead of the demand curve so the page has time to be crawled, indexed and to accumulate engagement before the searches arrive. Keep the same URL year after year and refresh the content — a page in its third year outranks a new one built in January.

Multi-location restaurants

Two or more rooms turns this into an architecture problem. The rules that hold up:

• One indexable location page per address, with unique copy, that location’s photos, its own hours, its own embedded map and its own reviews.
• One Google Business Profile per address, each with its own manager and its own posting cadence.
• A single brand-level menu page only if the menus are genuinely identical; otherwise a menu per location.
• Consistent NAP formatting across every profile, directory and page — down to the suite number and punctuation.
• Internal links from the homepage location switcher to every location page, and from each location page back to the relevant menu and booking pages.

AI search: how diners will find you next

Perplexity, ChatGPT Search and Google’s AI Overviews increasingly answer “best ramen near King Street” with three named recommendations and citations. Being cited depends on different signals than blue-link rankings:

Measuring what matters

Sessions are not a restaurant metric. Track these instead, monthly:

• Google Business Profile calls, direction requests and website clicks — the closest proxy to walk-ins.
• Reservation completions by source, with the booking platform’s referral data reconciled against analytics.
• Map pack ranking for your top ten dish and cuisine queries, measured from a grid of points around the restaurant rather than one address.
• Review count and average rating trend, plotted month over month.
• Catering and private dining enquiries, tagged by landing page.
• Share of AI answers citing you for your core cuisine query in your city.

A realistic 90-day plan

1. Days 1 to 14: foundations Audit and correct the Business Profile, categories and attributes. Convert the PDF menu to HTML. Fix hours including holidays. Compress images and get mobile load under two seconds.
2. Days 15 to 30: structure Ship Restaurant, Menu and FAQ schema. Build or rewrite the private dining and catering pages. Add one-tap booking to every template.
3. Days 31 to 60: demand capture Publish dietary and neighbourhood pages, launch the review request system, start weekly Google Posts and a monthly photo upload cadence.
4. Days 61 to 90: expansion Build the next seasonal campaign page, pursue local food media coverage and directory listings, and start reporting on calls, directions and reservations rather than sessions.

Frequently asked questions