Don MacMillan, who heads North America partnerships at Deliverect, says the data moving through his company's platform is now "the single source of truth for how a restaurant appears across search and AI." The claim sits in an announcement published on 8 October of a partnership with Yext, and the route it describes does not pass through the restaurant's own website at any point.
The feed carries menus, availability and allergens
Deliverect connects restaurants to the channels they take orders through, covering carryout, delivery, catering and in-store fulfilment across more than 80,000 locations in over 70 countries by its own count, with more than 1,000 certified integrations and 30 million API calls a day. October's announcement wires that layer into the Yext Knowledge Graph: "The integration connects Deliverect's operational data layer, including menus, opening hours, item availability, and allergen information, directly to the Yext Knowledge Graph."
From there Yext syndicates. Stephan Giordani, product partnerships manager for North America at Yext, describes the point of it as "eliminating a layer of manual data entry for our food and beverage clients", with the two companies making sure "their restaurants are accurately and consistently discoverable in AI and traditional search." A change entered once at the restaurant reaches search engines, maps and assistants with nobody retyping it.
Yext has run this pipe for restaurants since 2019
In December 2019 Yext announced it would publish brand-verified answers for Subway across "150+ search platforms like Google, Alexa, Siri, and Facebook", covering more than 30,000 restaurants in the United States, Canada and Europe, and building custom location pages alongside the listings. The syndication half of this month's news is seven years old.
What changed is the source. Not a marketing team filling in fields, and not a page crawled on the crawler's schedule, but the system that already knows which items the kitchen has switched off.
Live availability is what a crawled page cannot supply
Opening hours and allergen notes change rarely enough that well-marked-up pages keep pace. Availability does not. An item sells out mid-service and everything published about it stays wrong until someone edits the page and a crawler returns. A feed drawn off the ordering stack carries that state at the moment it changes, because the change is recorded there first.
Restaurants without a Deliverect connection and a Yext subscription are not short of ways to describe a menu. They are short of a way to say that tonight's special has gone.
The ordering menu is tuned for order value
Deliverect launched menu agents in April that use live order data to rearrange digital menus, lifting top sellers and pushing weaker items down, and operators can aim them at goals such as a higher average order value or a greater share of high-margin items. That is a reasonable thing to ask of a menu built to take orders.
Describing what a kitchen serves is a different job. Neither company has said whether an agent-arranged menu is what travels to the Knowledge Graph, and it is a fair question to put to any vendor offering to hold the record assistants read: which version goes down the wire, the one optimised for conversion or the one that matches the card on the table?
The release never says schema
"Schema.org" and "JSON-LD" appear nowhere in it. What it describes is "structured operational data" moving into what Yext calls "the world's most comprehensive structured data platform for local businesses", whose "API-first architecture connects structured data to APIs, MCP servers, and generative interfaces." That is a commercial pipeline with named counterparties at the far end, and it is also a real capability: an MCP server is how an assistant asks a question rather than waits to be told.
Markup on a restaurant's own pages works on a different footing. Anything permitted to fetch a public URL can read it, and no agreement governs the reading.
The two are not alternatives, and a restaurant signing up to one should know which problem it has solved. An accurate description it controls is one thing; a live availability signal routed through two vendors is another. A diner asking an assistant what is gluten-free tonight relies on both, and only one of them stays readable if the contracts change.
