I have run pre-engagement audits on a stack of local restaurants this year - the Google Business Profile, the listings, the website, the ordering setup, the data layer - and the picture is remarkably consistent. Strong brands with loyal regulars and good food, sitting on top of a digital presence that is quietly leaking on every layer. Not broken in dramatic ways. Leaking in small, fixable, compounding ways that nobody on the floor has had time to close.
What changed in 2026 is where the leaks hurt. The decision about where to eat used to happen in a Google search and a map result. It now also happens inside an AI answer that never shows the customer a website at all. ChatGPT, Gemini, Perplexity, and Google AI Overviews are reading your listings, your reviews, and your menu data and deciding whether to put your name in front of someone before that person ever opens a map. If your data layer is messy, you are now invisible in two places instead of one. This is the working audit I actually run before I recommend anything, updated for how discovery works this year.
Why do restaurants lose covers before anyone walks in?
Restaurants lose covers at the top of the funnel, in the few seconds where someone is deciding whether you are even an option. The owner thinks of that moment as a reservation or a walk-in. The real decision happens earlier - in a search, a map result, a profile scan, an AI answer, or a website load - and the restaurant that loses it usually never finds out it happened. There is no bounced-reservation report for the person who read your three-year-old cover photo and picked the place down the street.
Every fix below maps to that moment. None of it requires a big budget. Most of it requires an afternoon and the discipline to do it once, properly, instead of half-doing it every quarter.
What does Google actually rank restaurants on in 2026?
Google ranks local restaurants on three things - relevance, distance, and prominence - but the definition of prominence changed, and that is where most restaurants are losing ground. Prominence is no longer just review count and links. It is real-world engagement: how often people tap Call, how fresh your photos are, how recently you posted, and whether your reviews keep arriving or went stagnant a year ago.
The specific signals that move the needle this year, in the order I check them:
- Open at the time of the search. "Business is open when the customer searches" is now a top-five local ranking factor. Wrong holiday or seasonal hours do not just annoy callers - they actively suppress your Map Pack position during the exact hours you are trying to fill seats.
- Posting cadence. Google treats Business Profile posts as a freshness signal. The 2026 cadence that holds rank is roughly two posts a week. Most restaurants post twice a year.
- Review text, not just stars. Google's AI now reads the words in your reviews to infer your "vibe" and match you to queries like "date night" or "good for a group." Reviews that name specific dishes and occasions decide which searches you show up for.
- Review recency. A steady trickle of new reviews outranks a bigger but stagnant pile. Forty-seven reviews with the last one from 2024 reads as a business that may have closed.
- Photo freshness. Google tracks when photos were last updated and surfaces them prominently. A cover shot from 2021 competes against venues refreshing monthly.
The fix for the whole cluster is operational, not technical: lock the profile down once (verified phone, hours matching the front door), then build a recurring thirty-minute weekly block to post, respond to every new review by name, and add a few fresh photos. The technical work is twenty minutes. The habit is the actual deliverable.
The review response cadence that reads as "run by someone who cares"
A restaurant with thirty-seven reviews and zero owner responses reads as unmanaged. The same thirty-seven with a specific, non-templated reply on each one - especially the critical ones - reads as run by a human who is paying attention. The reply does not need to be long. It needs to reference something real in the review so it does not read as a canned auto-response, which both Google and the next reader can spot instantly.
Why is a PDF menu costing you AI search visibility?
A PDF menu is invisible to the systems that now decide where people eat. Google cannot index it cleanly, it opens slowly on mobile, and crucially, AI assistants cannot read structured menu data out of it. When someone asks an AI "where can I get good cacio e pepe near me," the restaurants with machine-readable menu data are the candidates. The PDF is not in the running.
The fix is to rebuild the menu as an HTML page on your own domain with proper heading structure and schema markup. Restaurant schema follows a clean tree that both Google rich results and AI answers read directly:
RESTAURANT SCHEMA TREE (JSON-LD)
-----------------------------------------
Restaurant
-> name, address, telephone (your NAP, identical
to GBP and listings)
-> servesCuisine (e.g. "Italian")
-> openingHoursSpecification (day-by-day, precise)
-> hasMenu
-> Menu
-> hasMenuSection
-> MenuSection ("Pasta")
-> hasMenuItem
-> MenuItem
-> name
-> description
-> offers (price,
priceCurrency)
-----------------------------------------
Format: JSON-LD (Google's preferred format)
Required minimum: Restaurant name + address;
MenuItem name + offers
Add FAQPage schema to the homepage and any
location page - it feeds AI answers directly.This matters more than it used to because rich results take roughly 58 percent of clicks versus 41 percent for standard blue links, and because AI search now answers menu, hours, and booking questions without sending the customer to a site at all. Structured data is how you stay in that answer. The added operational win: once the menu is an HTML page, updating a price is a CMS edit, not a call to a designer for a new PDF.
How do AI assistants decide which restaurant to recommend?
AI assistants assemble a recommendation from your listings and your reviews across the web, not from one source - which means consistency across platforms is now a ranking input, not just hygiene. Each assistant leans on a different spine: ChatGPT pulls heavily from Bing-indexed pages plus Yelp, Foursquare, and local guides; Gemini has a direct line into Google's local data; Perplexity leans on cited web sources and niche directories. If your name, address, phone, and hours disagree across Google, Apple Maps, Yelp, Bing, and the data aggregators, every one of those systems treats the conflict as a trust problem and quietly down-weights you.
The stakes are concrete. Roughly 40 percent of local business queries now trigger an AI Overview, and AI-cited recommendations are far scarcer real estate than the map pack. SOCi's 2026 visibility data puts it starkly:
| Surface | Share of locations that get recommended | What it rewards |
|---|---|---|
| Google local 3-pack | ~35.9% | GBP accuracy, proximity, reviews |
| Perplexity | ~7.4% | Cited web sources, niche directories |
| ChatGPT | ~1.2% | Bing index, Yelp/Foursquare, structured data |
The takeaway is not "chase ChatGPT." It is that the same three moves - clean consistent listings, structured data on the site, and review text that describes real experiences - feed all of these surfaces at once. You do not optimize for each AI separately. You fix the source data they all read.
Your listings are no longer directory entries that help people find you. They are the source data the AI uses to decide whether you are an option at all. Inconsistent data does not just lower your rank. It removes you from the answer.
What is the third-party platform actually taking from you?
The delivery and booking platforms take a percentage you can see and a customer relationship you cannot, and the second one costs far more over time. The visible cost is steep enough: DoorDash's tiered plans run 15, 25, and 30 percent commission for Basic, Plus, and Premier delivery; Uber Eats reaches up to 30 percent on delivery and, after its March 2026 marketplace-fee change, charges 7 percent on pickup with validated in-store pricing or 10 percent without. Once you add payment processing and the smaller deductions, the real all-in rate for most restaurants lands between 30 and 40 percent of the order.
The invisible cost is the one that compounds. The platform processes the order and hands you a ticket. It keeps the customer's contact information, the order history, and the relationship. When that customer reorders, they reorder through the platform, and the platform takes its cut again. You rented the transaction. They bought the customer.
The counter-move is to build a channel you own, in this order:
- Capture at the point of contact. A first-party ordering flow if you can, or at minimum a post-transaction capture - a table QR code to a short preference form, a checkout email field, a post-visit text. The goal is the email or mobile number landing in a list you control.
- Build the list deliberately. An email or SMS list of 1,500 people who have actually eaten there is an asset. Zero is a dependency on platforms that can change their algorithm or their fees whenever they like. From scratch, consistent capture gets you to 1,500 in about six months.
- Mark up your recurring assets. Happy hour, trivia night, seasonal specials - put them on a permanent page with Event schema. That is structured content Google and AI answers can surface, and most restaurants leave it completely invisible.
What is the order of operations?
The roadmap is three phases, run in sequence: stop the bleeding, build the assets, then own the data. Doing them out of order wastes money - there is no point driving paid traffic to a site that bounces, or building a list before the profile that feeds it is accurate. Here is the exact triage I run when a restaurant has limited time and budget:
- GBP accuracy and photos. These affect every impression before your site is even visited, and they are fixable in a single afternoon. Verify phone and hours, refresh the photo set, start the weekly posting block.
- Website speed and mobile experience. A slow site bleeds every traffic source, not just search. Run PageSpeed Insights on the homepage and menu page; the bar to clear is Largest Contentful Paint at 2.5 seconds or less, Interaction to Next Paint at 200 milliseconds or less, and Cumulative Layout Shift under 0.1, measured on mobile, since mobile is the primary ranking signal in 2026.
- Listings consistency and on-site schema. Reconcile NAP across Google, Apple Maps, Yelp, Bing, and the aggregators, then add Restaurant, Menu, and FAQPage schema to the site. This is what feeds both Google rich results and the AI answers.
- Review cadence and list-building. These are habits, not one-time fixes, so start them early and let them compound. Respond to every review; capture every customer you can.
- HTML menu and event schema. High value, lower urgency than the items above, but the move that takes you from indexed to recommended.
A restaurant that clears all of this in a quarter typically sees it in Map Pack position and foot traffic within sixty days. The work is not dramatic. It is steady infrastructure that compounds, which is exactly why most restaurants never get to it. If you want this run against your own location, that is the first conversation I have with every restaurant I work with, and the first pass is at no cost.
Common questions
Do I really need an HTML menu, or is a PDF fine if it is on my site?
You need the HTML menu. A PDF cannot be cleanly indexed by Google or read by AI assistants for structured menu data, so a PDF menu opts you out of menu-level search and AI recommendations entirely, no matter where it is hosted.
How do I get my restaurant recommended by ChatGPT or Perplexity?
Fix the source data all of them read: consistent NAP across Google, Apple Maps, Yelp, Bing, and aggregators; Restaurant and FAQPage schema on your site; and reviews whose text describes specific dishes and occasions. There is no separate ChatGPT setting - the assistants assemble answers from those same signals.
Is it worth leaving DoorDash and Uber Eats to avoid the fees?
Usually not entirely - they are a discovery channel. The move is to add a first-party ordering and capture path alongside them so repeat customers shift to the channel you own, where the all-in cost is a fraction of the 30 to 40 percent the platforms take.
How fast should my restaurant website load?
Aim for Largest Contentful Paint of 2.5 seconds or less on mobile, with Interaction to Next Paint under 200 milliseconds and Cumulative Layout Shift under 0.1. If your homepage or menu page scores below 75 on mobile PageSpeed Insights, you are losing visitors before the page finishes painting.
