Local Search Patents Vintage
Bill Slawski analyzed numerous patents related to how search engines handle geographically-bound queries. These patents reveal how Google determines local relevance, ranks local results, and understands the geographic intent behind queries.
Local Search Architecture
Location Sensitivity in Local Search
One of Bill's earliest local search patent analyses (2006) revealed that Google treats different types of local queries with different proximity thresholds.
Location Sensitivity by Query Type
| Query Type | Example | Distance Sensitivity |
|---|---|---|
| Food/Dining | "pizza" | Very tight — 5 miles or less |
| Automotive | "car dealer" | Wider — up to 50 miles |
| Professional Services | "lawyer" | Medium — 10-25 miles |
| Emergency | "hospital" | Extremely tight — nearest |
| Specialty Retail | "antique maps" | Very wide — regional |
The Sensitivity Model
The patent describes how a search engine might learn these sensitivity thresholds by analyzing user behavior — if users searching for "pizza" consistently click on results within 5 miles but ignore results 20 miles away, the system learns that pizza queries have high location sensitivity.
Source: Location Sensitivity in Google Local Search (2006)
Structured Information in Local Search
Bill analyzed how Google uses structured data from local businesses to improve local search results.
Types of Structured Local Information
| Data Source | What It Provides |
|---|---|
| Business listings | Name, address, phone, hours, categories |
| Structured data markup | Schema.org LocalBusiness, opening hours, menus |
| User reviews | Sentiment, ratings, specific attribute mentions |
| Photos | Visual confirmation, business type indicators |
| Menus/Product lists | Specific offerings that match detailed queries |
| Social profiles | Additional signals of legitimacy and activity |
How Google Glossary for Local Search Works
Bill documented a "local search glossary" from patents that defined key concepts:
- Entity reconciliation — Matching different references to the same business (e.g., "Joe's Pizza" on Yelp vs. Google Maps vs. the business website)
- Address normalization — Converting different address formats to a standard representation
- Category classification — Automatically categorizing businesses based on content analysis
- Operating hours extraction — Pulling business hours from unstructured text on websites
Address Completion and Geocoding
Patents describe systems for completing and interpreting addresses:
Address Processing Pipeline
Geographic Relevance Signals
Beyond simple proximity, patents describe multiple signals Google uses to determine geographic relevance:
Locally Prominent Semantic Features
A significant patent concept Bill highlighted was locally prominent semantic features — terms and concepts that are particularly important in specific geographic areas.
For example:
- "Lobster roll" is more semantically prominent in Maine than in Arizona
- "Surfing lessons" is more locally relevant in coastal cities
- "Snow removal" is more prominent in northern regions
What This Means for Local SEO
- Content that uses locally relevant terminology may receive a ranking boost in that geographic area
- Businesses should use region-specific language in their content
- The system learns what terms are locally prominent from analyzing content across a geographic region
Mobile Location History
Bill analyzed a 2018 patent on how Google uses mobile location history for local search.
Location History Data Points
| Data Point | How It Is Used |
|---|---|
| Frequently visited locations | Identifying home, work, and regular destinations |
| Travel patterns | Understanding commute routes and travel preferences |
| Visit duration | Determining how long users spend at businesses |
| Visit frequency | Identifying regular vs. one-time visits |
| Navigation usage | Where users ask for directions to |
Privacy and Local Signals
The patent describes how location history data is used to improve local search results by understanding individual user patterns. Google Maps tracks:
- Places you navigate to regularly
- Businesses you visit frequently
- Your typical geographic range
- Time patterns (when you search for restaurants vs. when you search for gas stations)
Source: Google's Mobile Location History (2018)
deCarta Mapping Patents
In 2012, Bill documented Google's acquisition of seven mobile location-based services patents from deCarta. These patents covered:
- Route-based search — Searching for businesses along a planned route
- Geofencing — Triggering actions when users enter/exit geographic zones
- Location-based advertising — Serving ads based on real-time location
- Spatial indexing — Efficient storage and retrieval of geographic data
Source: Google Scores 7 Mobile Location-Based Services Patents from deCarta (2012)
Key Takeaways
- Different query types have different geographic radii — Google applies different distance thresholds based on what you are searching for.
- Locally prominent terms boost rankings — Using language that is semantically relevant to a specific area provides a ranking advantage.
- Structured data is essential for local — Clean, complete business information in structured formats feeds directly into local ranking.
- Location history personalizes local results — Your past locations and behavior shape the local results you see.
- Address normalization matters — Consistent NAP (Name, Address, Phone) across the web helps Google's entity reconciliation.
- Mobile and local are deeply intertwined — Most local search patents originate from mobile device contexts.