The Leaked Binary File That Has Everyone Focused on the Wrong Thing
A binary file leaked last month, and the local SEO community lost its collective mind over the number 72. That's how many ranking signals Search Engine Land documented inside Geostore, the system Google uses to represent geographic entities. The coverage focused on the signals themselves: popularity metrics, web references, Knowledge Graph connections.
The signals matter. But if you're running marketing for a multi-location brand or advising one, the signals are the wrong place to start. The architecture underneath them changes what local visibility actually means, and it changes how you should allocate budget.
Your Listing Isn't the Entity
Here's the mental model shift that matters for planning: the Google Business Profile you edit is an interface. The entity Google maintains internally is something else entirely.
Resoneo's technical analysis of the leaked Geostore data makes this distinction explicit. Google builds a canonical representation of each place by combining data from 793 different providers. Your business name might come from one source, your phone number from another, your hours from a third. The system reconciles conflicts, assigns trust scores to each source, and produces a single entity that connects to the Knowledge Graph.
For B2B marketers managing distributed sales teams or franchise networks, this has immediate implications. Your GBP edits compete with other data sources. If your NAP data is inconsistent across directories, aggregators, and your own website, you're not just creating confusion for customers. You're lowering your trust score in a system that decides which source to believe.
The math here isn't complicated: consistent data across sources increases the probability that Google treats your claimed information as canonical. Inconsistent data means Google might pull your hours from Yelp, your phone number from a data aggregator, and your address from a two-year-old citation you forgot existed.
The Web Bridge Changes the Equation
The leaked architecture reveals something that should reshape how you think about local and organic as separate channels: they aren't separate.
Google connects geographic entities to the web through a system called Webref. When a document on the web mentions your business, that mention becomes evidence that feeds into the entity's profile. The Geostore data shows explicit connections between local entities and Knowledge Graph machine IDs, which themselves connect to web documents.
This means your content strategy and your local strategy are the same strategy. A location page that earns links and mentions isn't just helping organic rankings. It's feeding evidence into the entity that determines Maps visibility. A PR mention that references your business by name and location isn't just brand awareness. It's a signal that affects how Google understands your place in the world.
For budget allocation, this collapses what many organizations treat as separate line items. The question isn't "how much for local SEO and how much for content?" The question is "how do we create evidence that Google can use to build a more complete, more trusted entity?"
72 Signals, Zero Weights
The 72 ranking signals in Geostore include popularity metrics, prominence scores, web references, and semantic category data. Resoneo's analysis catalogued them in detail: signals for establishment popularity, signals for web presence, signals for category relevance.
What the leak doesn't include: the weights.
This is the part that should temper any tactical excitement. Knowing that Google uses a "popularity" signal doesn't tell you whether it's worth 2% or 20% of the final ranking. Knowing that web references matter doesn't tell you whether ten mentions from local blogs outweigh one mention from a national publication.
The practical response isn't to chase all 72 signals. It's to focus on the signals you can actually influence with measurable effort, then run experiments to see what moves the needle for your specific category and geography.
For most B2B organizations with physical locations, that means three areas: review velocity and sentiment, NAP consistency across the data ecosystem, and web content that creates entity-level evidence. Everything else is optimization at the margins until you've nailed those three.
AI Mode Changes the Visibility Math
Here's where the architecture story gets uncomfortable for anyone planning 2027 budgets.
Google's "Ask Maps" feature, powered by Gemini, is rolling out in the US and India. Instead of showing a list of three businesses, it often shows one. A user asks "find me a plumber who handles commercial properties and has weekend availability," and the AI returns a single recommendation.
The local pack was already winner-take-most. AI mode is winner-take-all.
This shifts the ROI calculation for local investment. If you're position three in a traditional local pack, you're still getting clicks. If you're not the AI's top recommendation, you're invisible for that query. The gap between first and second place widens from "fewer clicks" to "no clicks."
The Resoneo analysis notes that Google announced this AI wave on August 6, 2026, and the underlying infrastructure was already visible in the code. The system is designed to understand conversational queries and match them to entities that have complete, semantically rich profiles.
For planning purposes, this means the cost of an incomplete GBP just went up. Missing service descriptions, sparse attributes, thin review content: these aren't just missed optimization opportunities. They're reasons the AI might not recommend you at all.
The Pilot You Should Run This Quarter
If you're managing local visibility for multiple locations, here's a 30-day test worth running.
Pick five locations with similar baseline performance. For three of them, invest in completing every available GBP attribute, responding to every review within 24 hours, and publishing one piece of location-specific web content that mentions the business name, address, and primary service category. Leave two as controls.
Measure Maps impressions, direction requests, and phone calls at day 30 and day 60. The lag matters because entity updates don't propagate instantly.
The hypothesis: locations with complete semantic profiles and active engagement will show measurable lift in AI-influenced queries, which you can identify by looking at impression sources in GBP insights.
If the lift is real, you have a case for scaling the investment. If it's not, you've spent a month on hygiene work that needed doing anyway.
What This Means for 2027 Planning
The 72 signals are interesting. The architecture is actionable.
Google is building a system where your business exists as an entity connected to the web, scored by multiple ranking systems, and increasingly surfaced through AI that picks winners rather than presenting options. The organizations that win in this environment will be the ones that treat local visibility as an entity-completeness problem, not a listings-management problem.
That means budget for data consistency across the entire ecosystem, not just GBP. It means content strategy that creates web evidence for your locations, not just your brand. And it means measurement systems that can detect whether AI-mode queries are finding you or skipping you entirely.
The signals will keep changing. The architecture tells you where to build.
