# How search ranking works in food delivery apps

> A restaurant does not have a position in a delivery app. It has a different position at every point on the map, because ranking happens inside the set of restaurants that can deliver to the customer's address. The same brand can be third in one district and absent two kilometres away, with nothing changed on the restaurant's side. Platforms do not publish their ranking formulas, so the only number a restaurant can manage against is the position real customers see at real addresses.

Full page: https://kitchain.co/delivery-app-search-ranking/

## What is observable

Ranking formulas are proprietary and change without notice. Any claim about exact weights is speculation. What can be observed from the customer side:

Platforms do not publish ranking weights, but they do confirm the shape. Uber Eats states in its own merchant help that stores are ranked by an automatic algorithm and that every individual eater opening the app sees an entirely different ranking of stores from any other eater (Uber Help, checked 26 August 2026: https://help.uber.com/merchants-and-restaurants/article/how-does-uber-decide-rankings-in-the-uber-eats-app?nodeId=5ba17534-f0d7-4e13-90c3-d98f83989341). That is the clearest official confirmation available that a single published rank does not exist.

- **Location dominates.** The candidate set is restaurants deliverable to that address, ordered with distance and quoted delivery time weighing heavily. This is why position is district specific rather than city wide.
- **Open state matters.** Closed and paused stores fall or disappear from listings, so downtime and low visibility are entangled: a store with frequent outages accumulates absence from the feed.
- **Paid placement is mixed into the same surface.** Platforms sell sponsored positions and in app advertising, and the sponsored results share the screen with organic ones. What a customer sees is the blend, which is also what the restaurant is competing against.
- **Promotions change surfaces.** An active offer moves a listing into promotional carousels and filtered views that many customers browse instead of plain search.
- **Performance signals plausibly contribute.** Conversion, ratings and cancellations are commonly cited by platforms as inputs. Treat the direction as likely and the magnitude as unknown.

## The consequence: a rank surface, not a rank

A chain has a rank surface across the city. Measuring from the head office address, or reading the platform's own campaign report, samples one point of that surface and presents it as the answer. Useful measurement takes real points across the districts the brand cares about, records the customer facing listing at each, and repeats, so a change becomes a dated trend rather than an anecdote. Absence is part of the surface: a district where the brand never appears, while delivering there, is a silent loss no internal report shows.

## The personalisation caveat

Positions taken from the public, signed out view are reproducible and comparable over time. A signed in customer with order history may see a personalised ordering. Any product claiming to reproduce every individual customer's feed is overclaiming. The honest object of measurement is the public baseline and its movement, which is also the layer restaurants and platforms can discuss with each other.

## Why independence matters here

Restaurants buy visibility through commissions, CPC campaigns and funded discounts, and the platform selling that visibility is also the party reporting whether it worked. An outside record of positions across districts, before and during a campaign, is the only evaluation of that spend that does not come from the seller.

Measured context on the entanglement of downtime and ranking: in July 2026 UAE listings lost 1.68 percent of stated trading hours and 64 percent of unexpected outages began between 22:00 and 04:00, hours belonging to stores trading past midnight. A ranking report without store state at capture time hands those hours to the algorithm as if the algorithm caused them. Source: https://kitchain.co/uae-delivery-downtime-report/

Last updated 26 August 2026. Kitchain Visibility measures customer facing positions from map points saved per account, at set times of day in each city's own timezone, and records which competitor names appear above the brand at the same point: https://kitchain.co/ai/products/visibility.md. Category overview: https://kitchain.co/visibility-monitoring/