Leaked Google Maps data points to 72 Geostore ranking signals behind local search

What the recovered data contains

An examination of recovered Google Maps material, reported by Search Engine Land, offers a rare view of the infrastructure that likely sits beneath local search. According to the analysis, researchers obtained a binary exposing a non-public Geostore scope, then cross-referenced it with Maps protocols, captured network traffic, the web index, mobile services, style tables, on-device components and material from Google's 2024 leak.

The inventory that emerges is large: 72 Geostore ranking signals, 793 data source providers and 446 local search intent types, along with 50,998 Mapcore styles, 12,936 label styles and 10,936 searchable Geostore declarations. A separate on-device scorer is described as using eight signals across 13 tiers.

Oyster Rank is a vocabulary, not a formula

The 72 signals belong to an internal ranking vocabulary the analysis names Oyster Rank. That vocabulary operates inside a wider pipeline covering query understanding, semantic matching, candidate generation, geographic and quality evaluation, and reranking. Twenty-five of the 72 signals are flagged as deprecated, one of several reasons Search Engine Land warns against treating the list as 72 tactics or a confirmed Maps algorithm.

Google has not publicly confirmed any of the counts or the architecture described, and the analysis contains no coefficients, weights or causal ranking model.

Local listings modelled as entities

The central claim for practitioners is representational: Google appears to model each local listing as a geographic Feature rather than as the set of fields in a Google Business Profile. A Feature can carry identity, geometry, source information, websites, brand relationships, Knowledge Graph references, concepts and ranking data. A webref layer links documents to entities, which lets Google corroborate what a place is from evidence scattered across its systems.

Under that model, the profile is one visible surface of a much deeper object, and the ranking question becomes whether the wider evidence about the business agrees with it.

What this means for local SEO work

The analysis does not argue for abandoning Business Profile optimization; accurate core information still anchors the entity. The shift is toward consistency and completeness everywhere the business appears:

• Identity details — name, address, phone, website and attributes — matching across sources
• Website pages that state explicitly what the business offers, where it operates and which services belong to which location
• Accurate brand relationships and external references that reinforce the same picture
• Semantic completeness, with important facts written out plainly rather than implied through marketing copy

A dev.to write-up of the findings frames the takeaway as an audit exercise: compare the profile against location pages, service descriptions, contact details and third-party references, correct contradictions first, then fill the gaps that leave the entity ambiguous.

Limits of the evidence

Search Engine Land describes the work as reverse-engineered analysis rather than an official disclosure. It does not show how signals interact, which factors decide a given query or market, or whether every listed signal is still active. The most defensible reading is a map of internal concepts and possible evaluation stages, not a guaranteed route to better Maps positions.

Why it matters

Local SEO has long centred on filling Business Profile fields correctly. If the recovered data reflects production systems, visibility depends instead on how coherently an entity is documented across the profile, the website and connected references — a broader and harder-to-game surface than field-by-field editing. The timing sharpens the point: as Maps moves toward AI-assisted conversational experiences, ranking systems must reason about entities and their relationships rather than retrieve a matching category. Businesses whose public information answers basic questions unambiguously — what they offer, where, and under which brand — are better positioned for both today's Maps and the AI interfaces now mediating local discovery.